Question 4080 Super died

[pacha298]

[Moderator edit to break up solid paragraph of text.]

Video Good afternoon all, about 2 weeks ago i built my newest pc and all was perfect until i left my pc on overnight to update a few games,

i woke up about 6 or 7 hours later to find that my pc shut off, i thought the power went out overnight because it was quite windy but it didnt, only the ram was lit up inside of my case.

Thinking it was the psu I went out and bought a brand new corsair rm1000x and took it to a tech professional to test the old (new deepcool px850g) psu and get a second opinion.

Come to find out the psu was working as it should and there was nothing wrong with it. He tried to boot the pc again after testing it and still nothing, so he swapped the 4080 super with his own card and my pc booted up as if nothing was wrong, went through all the debug lights.

The deepcool px850g had a direct 12vhpwr connector from psu to gpu which i was using because it didnt have enough pcie cables to use the nvidia adapter without daisy chaining.

Im not an expert electrician so im not very knowledgeable about amps, watts, volts and what have you.

Did the 12vhpwr cable screw me over? Did i get a bad card? I've tried asking on reddit and all i got was belittling replies and no true answer.

I liked using the 12vhpwr connector because the nvidia adapter looks stupid and it was much easier for cable management and neatness.

The corsair rm1000x doesn't have a 12vhpwr unfortunately :/ I just shipped the dead gpu back to newegg today and should expect the new one sometime this week.

Full part list
Aorus B650 elite ax ice
Tforce delta ddr5
4080 super aero
deepcool px850g
Ryzen 7 7800x3d
samsung 990 pro

 

[Crowii]

a bit of a nothing-burger answer, but I don't really think you should expect a conclusive answer for this post. It boils down to: "Why did my GPU die, and will it happen again?"

To see if anything electric, and relevant to the 12vhpwr connector, has fried something requires physical inspection/probing. You'll truly find out if something is wrong with the rest of the system if another GPU dies.

I'd take a close look at the 12vhpwr cable that came with the PSU and look for any anomalies; it is a notoriously fickle (and personal opinion: dumb and unrealistic) standard at this point. From the slightest bend in the plastic, to a molding glob, to an inconsistent angle in the fingers on the pinouts etc. If anything looks wrong, I'd contact deepcool and see if they think it's a defect, and if they'd be willing to send a new cable. But in general it should almost always be better than the octopus adapter Nvidia ships.

Likewise make sure you really wrench the cable in both on the PSU and GPU side, I've had multiple build where I've gotten a clear click but is still very visibly a crooked connection.

 

[Ralston18]

Did the tech professional do any testing with or without the 12vhpwr?

Even though the 12vhpwr connector provides a different look and ( perhaps more practically) provides improved cable management the system should be rebuilt without the 12vhpwr.

Consider that the 12vhpwr may be defective and/or caused damage as a result.

Redo the build without the 12vhpwr connector.

= = = =

Note: this PSU?

https://www.amazon.com/Corsair-RM1000x-Fully-Modular-Supply/dp/B08R5PH1VY?th=1

Modular and only the cables that come with any given PSU should be used.

 

[pacha298]

Did the tech professional do any testing with or without the 12vhpwr?

Even though the 12vhpwr connector provides a different look and ( perhaps more practically) provides improved cable management the system should be rebuilt without the 12vhpwr.

Consider that the 12vhpwr may be defective and/or caused damage as a result.

Redo the build without the 12vhpwr connector.

= = = =

Note: this PSU?

https://www.amazon.com/Corsair-RM1000x-Fully-Modular-Supply/dp/B08R5PH1VY?th=1

Modular and only the cables that come with any given PSU should be used.

yes, he also tried the card with a different psu and the nvidia supplied pcie adapter, nothing worked until he swapped the card for his own 6900xt. also the psu you linked is the one i bought most recently, however the problem occured with the deepcool psu

 

[pacha298]

a bit of a nothing-burger answer, but I don't really think you should expect a conclusive answer for this post. It boils down to: "Why did my GPU die, and will it happen again?"

To see if anything electric, and relevant to the 12vhpwr connector, has fried something requires physical inspection/probing. You'll truly find out if something is wrong with the rest of the system if another GPU dies.

I'd take a close look at the 12vhpwr cable that came with the PSU and look for any anomalies; it is a notoriously fickle (and personal opinion: dumb and unrealistic) standard at this point. From the slightest bend in the plastic, to a molding glob, to an inconsistent angle in the fingers on the pinouts etc. If anything looks wrong, I'd contact deepcool and see if they think it's a defect, and if they'd be willing to send a new cable. But in general it should almost always be better than the octopus adapter Nvidia ships.

Likewise make sure you really wrench the cable in both on the PSU and GPU side, I've had multiple build where I've gotten a clear click but is still very visibly a crooked connection.

I understand, after further inspection of the 12vhpwr cable i've noticed that on the psu side, there is slight play where the cable can come undone a bit without pushing the tab down. Perhaps it could've came loose somehow on its own? After i finished cable management with this build i checked all wiring to be properly secured and inserted.

 

[Crowii]

I understand, after further inspection of the 12vhpwr cable i've noticed that on the psu side, there is slight play where the cable can come undone a bit without pushing the tab down. Perhaps it could've came loose somehow on its own? After i finished cable management with this build i checked all wiring to be properly secured and inserted.

It's not impossible that it might've loosened, especially after maybe getting tightened/strained from cable management? Or even just a good knock from being put on the floor or another table. I think even when being careful, fully checking the PSU side of the connection is easily overlooked just due to being crowded. It really just isn't a good connector in any way, shape, thought or form - so a repeat reminder that nothing can really be said for sure without handling the system in person.

Like @Ralston18 said, I'd try with the new PSU and the octopus adapter with the new GPU, just in case the old Deepcool PSU's 12vhpwr has been damaged.

Should you still be willing to experiment with the Deepcool PSU, then I'd still contact their CS and see if they'd be willing to send a new cable after explaining your situation. And then just hope it's a faulty cable, and not the whole 12vhpwr delivery that is damaged and/or faulty.

 

[pacha298]

It's not impossible that it might've loosened, especially after maybe getting tightened/strained from cable management? Or even just a good knock from being put on the floor or another table. I think even when being careful, fully checking the PSU side of the connection is easily overlooked just due to being crowded. It really just isn't a good connector in any way, shape, thought or form - so a repeat reminder that nothing can really be said for sure without handling the system in person.

Like @Ralston18 said, I'd try with the new PSU and the octopus adapter with the new GPU, just in case the old Deepcool PSU's 12vhpwr has been damaged.

Should you still be willing to experiment with the Deepcool PSU, then I'd still contact their CS and see if they'd be willing to send a new cable after explaining your situation. And then just hope it's a faulty cable, and not the whole 12vhpwr delivery that is damaged and/or faulty.

I will give it a shot, in the event that everything is fine with the new gpu and octopus adapter would you recommend switching back over to a 12vhpwr cable? I was looking into maybe getting the corsair supported cablemod set with a 90° adapter to reduce strain and bending on the 12vhpwr connector

 

[Crowii]

I will give it a shot, in the event that everything is fine with the new gpu and octopus adapter would you recommend switching back over to a 12vhpwr cable? I was looking into maybe getting the corsair supported cablemod set with a 90° adapter to reduce strain and bending on the 12vhpwr connector

Recommend would be a strong word if we're talking going back to the old Deepcool PSU since it's not unlikely it just killed a GPU. I can't in good faith recommend that over a forum post. If that's what you wish, then all I can recommend is at least get a replacement cable.

If going back to the cleaner 12vhpwr look without the adapter then I'd look at returning the corsair PSU, or selling it 2nd hand if not possible, to subsidize the cost of getting a new one that supports it.

 

[pacha298]

Recommend would be a strong word if we're talking going back to the old Deepcool PSU since it's not unlikely it just killed a GPU. I can't in good faith recommend that over a forum post. If that's what you wish, then all I can recommend is at least get a replacement cable.

If going back to the cleaner 12vhpwr look without the adapter then I'd look at returning the corsair PSU, or selling it 2nd hand if not possible, to subsidize the cost of getting a new one that supports it.

Its extremely doubtful that i'll go back to the deepcool psu. If anything i would return the corsair psu and get a pcie 5.0 compatible psu. The only reason i bring up going back to the 12vhpwr one is the 90° adapter, i know a lot of people were having issues with their graphics cards because of the cable bending too sharp. Also i hate how chunky the nvidia adapter is haha.

 

[boju]

Can get 12vhpwr adapter from Corsair.

https://www.corsair.com/us/en/p/pc-...5-0-12vhpwr-type-4-psu-power-cable-cp-8920284

When leaving computers on overnight i think it's best to adjust power plan to never turn off hard disks and never turn off display, just turn off monitor manually. These settings are in the advanced section for the power plan you have set. Imo, i dislike the idea of sleep also, so many people have problems with it, i don't have reasons to use it so i avoid using that function. Not saying there's a connection to your issues with sleep or power options, it's just a suggestion from my own experience how i run computers. While im still sharing lol, i never let graphics cards power save either, always maximum performance, in the belief that letting it power fluctuate does more harm than good. But that's just me, never had issues running graphics cards like that over the many years of use playing games. Im either completely factually wrong about that or possibly onto something, again, just speaking from my own experience.

 

[Crowii]

Its extremely doubtful that i'll go back to the deepcool psu. If anything i would return the corsair psu and get a pcie 5.0 compatible psu. The only reason i bring up going back to the 12vhpwr one is the 90° adapter, i know a lot of people were having issues with their graphics cards because of the cable bending too sharp. Also i hate how chunky the nvidia adapter is haha.

Sadly these 90° adapters also haven't proven their track record of being a reliable fix, circling back to 12vhpwr just being a shoddy standard, even for 3rd parties trying to follow the guidelines to dot. It's probably better than nothing, and better than the abomination 4 split octopus Nvidia thought was okay, but still.

When leaving computers on overnight i think it's best to adjust power plan to never turn off hard disks and never turn off display, just turn off monitor manually. These settings are in the advanced section for the power plan you have set. Imo, i dislike the idea of sleep also, so many people have problems with it, i don't have reasons to use it so i avoid using that function. Not saying there's a connection to your issues with sleep or power options, it's just a suggestion from my own experience how i run computers. While im still sharing lol, i never let graphics cards power save either, always maximum performance, in the belief that letting it power fluctuate does more harm than good. But that's just me, never had issues running graphics cards like that over the many years of use playing games. Im either completely factually wrong about that or possibly onto something, again, just speaking from my own experience.

If this was a case of extremely long term degradation of hardware then I'd call this somewhat relevant, but given that this a straight up fried recent GPU, I'd say otherwise. This is entering 24/7 server load territory and imo isn't really relevant for consumer hardware. If heat cycles (which is what you're really referring to with power cycles) from varied loads causes your hardware to fail before it's simply considered so old that you might as well replace it (for reference, that'd be like running a repasted GTX 5xx series at the earliest or older these days).

If there was and actual relevant issue on a conflict between sleep/power states causing either an oversight in firmware and/or improper behavior in any sleep states, then that'd be an easy search to hit and would be brought in here.

And if you're on Windows, then you're still using a sleep state of some kind unless disabled in registry or by some way in bios. There's S0-3 (S1-4 in documentation), and simply clicking "Shut Down" isn't even S0. Being practically a spectrum at this point with 4 different sleep stages, there really isn't any weight behind not liking "sleep state" without some good research behind it. And to keep vaguely relevant to this thread, for consumer hardware and use case, not 24/7 use cases.

 

[pacha298]

Can get 12vhpwr adapter from Corsair.

https://www.corsair.com/us/en/p/pc-...5-0-12vhpwr-type-4-psu-power-cable-cp-8920284

When leaving computers on overnight i think it's best to adjust power plan to never turn off hard disks and never turn off display, just turn off monitor manually. These settings are in the advanced section for the power plan you have set. Imo, i dislike the idea of sleep also, so many people have problems with it, i don't have reasons to use it so i avoid using that function. Not saying there's a connection to your issues with sleep or power options, it's just a suggestion from my own experience how i run computers. While im still sharing lol, i never let graphics cards power save either, always maximum performance, in the belief that letting it power fluctuate does more harm than good. But that's just me, never had issues running graphics cards like that over the many years of use playing games. Im either completely factually wrong about that or possibly onto something, again, just speaking from my own experience.

I used to run my old pc for like 3 days straight and never had an issue, im pretty sure i switched the power settings and turned sleep mode off inside of the windows settings, might be related to my issue or not but on my second monitor my task bar is actually burned into the display so i can see it when the display is off lol, i dont remember that being there before

 

[pacha298]

Sadly these 90° adapters also haven't proven their track record of being a reliable fix, circling back to 12vhpwr just being a shoddy standard, even for 3rd parties trying to follow the guidelines to dot. It's probably better than nothing, and better than the abomination 4 split octopus Nvidia thought was okay, but still.

If this was a case of extremely long term degradation of hardware then I'd call this somewhat relevant, but given that this a straight up fried recent GPU, I'd say otherwise. This is entering 24/7 server load territory and imo isn't really relevant for consumer hardware. If heat cycles (which is what you're really referring to with power cycles) from varied loads causes your hardware to fail before it's simply considered so old that you might as well replace it (for reference, that'd be like running a repasted GTX 5xx series at the earliest or older these days).

If there was and actual relevant issue on a conflict between sleep/power states causing either an oversight in firmware and/or improper behavior in any sleep states, then that'd be an easy search to hit and would be brought in here.

And if you're on Windows, then you're still using a sleep state of some kind unless disabled in registry or by some way in bios. There's S0-3 (S1-4 in documentation), and simply clicking "Shut Down" isn't even S0. Being practically a spectrum at this point with 4 different sleep stages, there really isn't any weight behind not liking "sleep state" without some good research behind it. And to keep vaguely relevant to this thread, for consumer hardware and use case, not 24/7 use cases.

Yeah, i feel like the 4xxx series launch was rushed and despite these blatant issues they refuse to fix it and let their ego get the best of them claiming nothing is wrong. If the same thing happens with this new card that newegg is sending out should i seek a warranty replacement through nvidia?

 

[Crowii]

Yeah, i feel like the 4xxx series launch was rushed and despite these blatant issues they refuse to fix it and let their ego get the best of them claiming nothing is wrong. If the same thing happens with this new card that newegg is sending out should i seek a warranty replacement through nvidia?

None of this responsibility is on Nvidia, sadly. The manufacturer bought the silicone from Nvidia, built a card and cooler around it to their rules, boxed it and sold it to Newegg. That means the faulty hardware agreement is either on Newegg or the manufacturer the card originates from.

For now I'd say this is hopefully a one-off for you, and the new card will work with the PSU discussion above. Any further issues can be brought up in a new post

 

[bill001g]

I suspect you are worrying about nothing. All the problems you hear about the 12vhpwr connector is the melting and that was mostly on 4090 that were running above 450 watts. There were also a bunch of cards where you could break the solder connections on the boards by pressing too hard....mostly because people were worried about the melting issue.

You do not see complaints that the cables somehow damaged other parts of the GPU. The cables are designed to provide the proper amount of power. one that was defective the video card might not boot but on a different cable or machine it would.

They will likely send the card to a repair place. In some ways it would be interesting to see what they found on your card but they will likely just be sending you a different card that someone else sent in that was repaired previously. There are a bunch of youtube channels with guys fixing video cards. It seem some fairly minor inexpensive thing failed that can kill the card. Although they make it look simple it take a lot of skill to replace a part you can barely see. It is good of newegg to take this back some video cards you must go get the warranty service from the card manufacture.