Melted PCIE power connectors - GPU broken?

[AWT]

Hi.

After having struggled with the resulting problems for over a week (and now finally not being able to boot any longer), I have found out why I keep getting monitor signal drops and failed POSTs with VGA light on.
As you can see in the following pictures, my PCIE power cables melted in my RTX 3080 (strangely I had no issues removing/replacing them).
https://imgur.com/a/VG8I03d
(Full specs in my profile/bio).

Since replacing the cable (I have a modular PSU) hasn't solved the problem, I wonder if the card might have been damaged.
Not only are some of the connectors now shorter than others on the damaged cable, but one or them also seems deformed, being kinda oval rather than round (even though all the cards pins are round).
I will see for sure when my new 7700XT arrives next week (as I'll need something to use the PC while the RTX goes to RMA).

I managed to get into UEFi once after replacing the cable, but as soon as I had the basic settings done and pressed save and restart, the boot process failed again (with the VGA light coming on indicating VGA issues). Mainboard also displays a d6 code, meaning "no console output device detected". Am I right to assume the GPU is defective given the indication on the Mainboard?

Thanks in advance for any advice.

[Illusion of Progress]

As far as I know, usually, when this happens, it's because of poor contact at point of connection, or some other factor (like stress or a sharp bend), either of which results in heat building up at the contacts that are good, and... it melts the plastic.

Usually, unless something else goes wrong, it doesn't necessarily mean the GPU incurred damage, but based on your symptoms, it does sound like system is failing to POST due to the video card.

Do you have onboard video you can test with?

[AWT]

Unfortunately not.
Update - I managed to boot the computer now, but when I enable Resize BAR, the white VGA light comes on and I get no picture until the Windows Welcome screen appears.
The card is also drawing a constant 100W in idle, with the fans running at full speed, and it is not detected in Geforce Experience (GPU Z recognizes the card though).
(I reset CMOS to get back into UEFI and disabled Resize BAR and enabled CSM; now I do get VGA signal before Windows is loaded and the VGA light no longer comes on, but the card still runs at full fan speed + 100W idle draw and still not detected by GeForce Experience).
Does this indicate the card is damaged? It seems like it to me, but maybe I forgot to check something (I'm quite happy it booted at all at this point, as the computer didn't boot at all for the last 3 days until I found out about the melted cable and replaced it - so I'm quite tired of troubleshooting by now). Thanks in advance for any advice.

[Illusion of Progress]

It's hard to say.

I've had some monitoring programs acting wonky lately myself.

Libre Hardware Monitor shows a lot of odd numbers (like "NaN" [not a number] and the infinity symbol, along with values like 13 GHz to 15 GHz) and core voltages of 1.5V (normally I see it stay around 1.25V and under).

Afterburner stopped reporting CPU frequency and GPU temperatures.

GPU-Z doesn't read a lot of things.

CPU-Z reads the CPU right but not the GPU.

Latest versions of all of them, by the way.

HwInfo64, Adrenalin (AMD drivers), and what little the Windows Task Manager shows is fine though. So I presume those prior programs just aren't right anymore.

I think this happened after a motherboard BIOS update within the last few months because nothing else I can think of changed on my system (besides Windows updates).

So odd readings on their own I wouldn't think much of. But combined with the other odd issues, maybe?

Something I overlooked but forgot to ask, but when you changed the cable that had damage on it, did you replace it with the same cable? This is important, because these cables aren't always compatible with others. Meaning using any random modular cable with some PSUs doesn't always work. Sometimes, it results in catastrophic damage when attempting to power on.

At this point I'd try removing everything with a connection, reconnecting it, ensure the metal contact points on the graphics card where the PCI Express cable plugs is in clean (since the prior one started melting), and if the cable isn't the proper one for the PSU, definitely try a matching PSU and cable.

No other system you can test the graphics card in either?

With luck, when you catch a melting connector early, usually the GPU, PSU, and system is fine. Hopefully that's the case here.

By the way, expect the 7700 XT to be a bit slower (in case you don't know). The 6800 XT or 7800 XT would be a closer match to your RTX 3080.

[_I_]

Postat inițial de Illusion of Progress:

It's hard to say.

I've had some monitoring programs acting wonky lately myself.

Libre Hardware Monitor shows a lot of odd numbers (like "NaN" [not a number] and the infinity symbol, along with values like 13 GHz to 15 GHz) and core voltages of 1.5V (normally I see it stay around 1.25V and under).

Afterburner stopped reporting CPU frequency and GPU temperatures.

this often happens when you run 2 monitoring softwares at the same time

they both ask the sensor at different times and get mixed up, both listening to the sensors output

[Illusion of Progress]

Postat inițial de _I_:

this often happens when you run 2 monitoring softwares at the same time

they both ask the sensor at different times and get mixed up, both listening to the sensors output

I don't have two running (at least as far as I know?), and I could even run two in the past and it never happened. I believe this occurred after I updated my motherboard BIOS (or some other Windows update change) but I don't run monitoring stuff often enough to have a smaller time window to know what exactly did it.

[Azuwi]

you can't replace a modular psu cable with another one, it will melt/fry your GPU. i learned this the hard way.

since your GPU is broken now, you will fry your new gpu too if plugged in.

you have to buy a new psu and a new graphic card, then your pc will work again.

[Phénomènes Mystiques]

Postat inițial de Azuwi:

you can't replace a modular psu cable with another one, it will melt/fry your GPU. i learned this the hard way.

since your GPU is broken now, you will fry your new gpu too if plugged in.

you have to buy a new psu and a new graphic card, then your pc will work again.

You can also get new cables for the specific PSU model from most brands but yes it has to be the specific cable to that specific model (no third-party cables!) for it to match the polarity (and tolerances) as some are reversed and that causes the issues of "frying".

[Jamebonds1]

Main reasons it melted port are:
Wrong cable type for the PSU manufacturer

Too loose connection

Bent cable in wrong place

Used adapter

And many reasons, but dmm can help you find issue.

[smokerob79]

I will be blunt.....this should not be happening on a 3080......good odds something is wrong with the card....my EVGA 3080 is in about 20 watts idle.......

[_I_]

should not but clearly does

anything with the 12pin hpwr connector has potential to melt if not connected securely and correctly
which is why the gpu and psu and cable mfgs recommend not to have any bends or flex in the cable for about 1.4in or 35cm from the connector

from seasonic
https://www.1fodiscount.com/ressources/site/document/product/brochure-commerciale-pour-seasonic-12vhpwr_10706.pdf

[Illusion of Progress]

The RTX 3080 doesn't use the 12VHPWR connector though. It uses two or three 8-pin connectors.

[_I_]

ok, the molex 6/6+2 connectors are much more forgiving, they will not fail, unless visibly not seated fully or even fail at all

3070 is 220w, it should use a 6+2 and 6pin
if you try hard enough you might be able to shove 4/4+4 eps in one, but the psu should short fault detect and not turn on

[Azuwi]

Postat inițial de The Incredible Magical Cupcake:

Postat inițial de Azuwi:

you can't replace a modular psu cable with another one, it will melt/fry your GPU. i learned this the hard way.

since your GPU is broken now, you will fry your new gpu too if plugged in.

you have to buy a new psu and a new graphic card, then your pc will work again.

You can also get new cables for the specific PSU model from most brands but yes it has to be the specific cable to that specific model (no third-party cables!) for it to match the polarity (and tolerances) as some are reversed and that causes the issues of "frying". :csd2smile:

In my case I had a fractal design psu and I used a corsair psu cable and it fried both gpu and motherboard if i remembered. :D

[Illusion of Progress]

Postat inițial de _I_:

ok, the molex 6/6+2 connectors are much more forgiving, they will not fail, unless visibly not seated fully or even fail at all

3070 is 220w, it should use a 6+2 and 6pin
if you try hard enough you might be able to shove 4/4+4 eps in one, but the psu should short fault detect and not turn on

RTX 3080, not 3070, and the thread starter showed a picture above of what the cable looks like. Cable looks a bit melted to me, but unless that was poorly connected all along and had poor contact, I'm not sure what would cause it here.

[AWT]

Update - after uninstalling the driver and windows auto-installing a driver, the card now runs at 27W idle as normal.
The only thing still worrying me is that when I enable Resize BAR, the monitor only receives signal once windows is loaded - so I'll have to reset CMOS again whenever I want to change some settings in the system. I think I will disable that feature.
What is also quite worrying is that the CPU instantly jumps to 90°C in a stress test after re-enabling core performance boost - even though none of them boosts to 4.1 GHz as it did before.

The cable I replaced it with was labeled PCIE, just like the one that melted (a bit worrying that the PSU manufacturer chose to add a spare incase one melted - seems like they were almost expecting this to happen).
The 3080 I have uses 2x8pin standard PCIE cables (as shown in the picture I shared in the original post that started the thread), which in my mind is a bit of a design flaw - the card uses up to 350W, so both PCIE power cables are basically running at max capability when the card is under full load. I guess the risk of cables melting wouldn't be as high if they used a 3rd connector/cable to lower the load bearing upon the cables at full power.
Maybe that 7700XT I ordered might be a good alternative given that it offers almost the same performance at 250W...