"Fixed" a shady GTX 1080 today. It can't be that easy, right?

[koralski]

I am guessing that a 1000W hairdryer won't do the job? Even if held for a long time?

I am starting to suspect its a bad connection to something. If the card warms up it doesn't crash. But if I take it out of the machine for a day and try and start a game it'll crash.

Exactly what happened the day after I posted the thread, woke up. Installed the GPU in my GF's machine. Ran Cyberpunk, and it crashed.

 

[eidairaman1]

Fair points, but a key aspect is effectiveness of application. Lower water content means the IPA potency will be higher and the cleaning effect will be greater. 90ish% is the sweet-spot of cost and usefulness. 70% will still get the job done, as would 50%. One just has to use more of it.


True. Care should be taken to avoid using 90%+ on certain plastics.

Yeah im trying to see if i can find a conditioner to restore the color

 

[koralski]

I've had OCCT run for an hour on the VRAM. No errors there.

Kombustor ran for an hour and zero issues there also. WTF?

Currently undervolted at -150 both at core and memory.

 

[Shrek]

Could try a cold spray on parts of the board.

Freeze Spray: Diagnosing Faulty Electronics

Freezer is used to test electronic equipment for and locate thermal intermittent components and systems, provide heat sink protection for heat sensitive components during soldering and desoldering, perform low-temperature testing of prototype circuitry and electromechanical systems, and locate...

www.techspray.com

'Compressed air' cans can be used for this by turning them upside down.

 

[koralski]

Could try a cold spray on parts of the board.

Freeze Spray: Diagnosing Faulty Electronics

Freezer is used to test electronic equipment for and locate thermal intermittent components and systems, provide heat sink protection for heat sensitive components during soldering and desoldering, perform low-temperature testing of prototype circuitry and electromechanical systems, and locate...

www.techspray.com

'Compressed air' cans can be used for this by turning them upside down.

Seems interesting, I would give it a try.

I replaced my GF's RX580 with the 1080, if crashing continues I will try this and torching it. If all else fails it'll make for a nice wall ornament. I remember drooling over the idea of owning this card years ago...

Could try a cold spray on parts of the board.

Freeze Spray: Diagnosing Faulty Electronics

Freezer is used to test electronic equipment for and locate thermal intermittent components and systems, provide heat sink protection for heat sensitive components during soldering and desoldering, perform low-temperature testing of prototype circuitry and electromechanical systems, and locate...

www.techspray.com

'Compressed air' cans can be used for this by turning them upside down.

Also I want to ask you, will an ordinary heat gun, like a DeWalt for an example work fine for this application? I am looking to rent one out, I don't see value in buying a heat gun for this card.

 

[Shrek]

$10 to rent, $20 to purchase...

 

[koralski]

$10 to rent, $20 to purchase...

4.60 USD to rent it, about 28 USD to buy a Black&Decker brand. I am in bumfuck nowhere, not in the US.

 

[Marcus L]

bumfuck nowhere

 

[koralski]

I don't wanna jinx it, but the card so far works. I can't believe it.

 

[koralski]

The card still works... WOW. -150 on core, -150 on memory and 90% power limit, just to be sure. It seems stable with stock clocks but I don't wanna cause further degradation.

 

[lexluthermiester]

The card still works... WOW. -150 on core, -150 on memory and 90% power limit, just to be sure. It seems stable with stock clocks but I don't wanna cause further degradation.

That is a sure-fire way to stabilize a card that is on the edge of stability.

 

[koralski]

That is a sure-fire way to stabilize a card that is on the edge of stability.

Is that an ironic comment, or are you serious? I really can't tell.

 

[lexluthermiester]

Is that an ironic comment, or are you serious? I really can't tell.

Totally serious. Lowering clocks and voltage is one of the many excellent ways to stabilize a GPU having stability issues. It's a universally understood and accepted solution. Why would you think that is a joke?

 

[Leooooo1061]

If the card works then great, although the die is pretty obviously degraded. Underclocking and undervolting totally helps stabilize a gpu; there's nothing wrong with it. I strongly suspect it is used for mining before... you don't get that much dust and crap inside a normal gamer's pc, although a few did come close in this thread: The Filthy, Rotten, Nasty, Helpdesk-Nightmare picture clubhouse. Some of those pictures always give me a good laugh. However, reflowing might actually improve the situation somehow, as you mentioned tightening the backplate screws actually made the artifacting issue slightly better(although you shouldn't really do that. It might crack the die). I would STRONGLY suggest NOT using an oven though. It heats up the whole card and those resistors on the back might fall off. In addition to that you get a smell of burning crap inside your oven and a risk of melting the 8-pin pcie power connector. You should buy a heat gun and maybe watch some Youtube videos to get a better understanding of how to reflow stuff. Also getting some thermal pads and applying them to the VRAM might help, as artifacts are usually associated with them.

 

[AsRock]

I've had OCCT run for an hour on the VRAM. No errors there.

Kombustor ran for an hour and zero issues there also. WTF?

Currently undervolted at -150 both at core and memory.

But maybe when the GPU and ram is running at full pace the ram gets to hot maybe ?, maybe wrong pads on them ?. Or even a pad missing from some were on the board.

Better to check what the pad thickness is and make sure.

And if you must try a heatgun better to shield those caps as it will damage them and while at it see if the caps are bulging or not.

 

[Leooooo1061]

But maybe when the GPU and ram is running at full pace the ram gets to hot maybe ?, maybe wrong pads on them ?. Or even a pad missing from some were on the board.

Better to check what the pad thickness is and make sure.

And if you must try a heatgun better to shield those caps as it will damage them and while at it see if the caps are bulging or not.

Yeah that's probably a smart thing to do. Better use some Kapton tape to cover things up. Pay attention to the pcie power connector as well; I've melted one of those while soldering loose resistors on a GTX 750.

Also THIS is what you'll get if you put stuff in the oven:

Not sure if I posted this yet, but here we go.

That's on topic of people with questionable intelligence baking shit on every imaginable occasion. Poor GTX660 only had a malfunctioning PWM controller, before it was put in the torture oven.
View attachment 82927

If someone is offended by my post and wants to mention some examples of successful "baking", please move on. I am short-tempered.

 

[koralski]

But maybe when the GPU and ram is running at full pace the ram gets to hot maybe ?, maybe wrong pads on them ?. Or even a pad missing from some were on the board.

Better to check what the pad thickness is and make sure.

And if you must try a heatgun better to shield those caps as it will damage them and while at it see if the caps are bulging or not.

Yeah, I've been watching northwestrepair, he cuts off soda cans and makes a shield with that. Probably gonna do that if the card shits itself again.

I have calipers, I'll measure the thickness. Any ideas what pads to get, what brand?

If the card works then great, although the die is pretty obviously degraded. Underclocking and undervolting totally helps stabilize a gpu; there's nothing wrong with it. I strongly suspect it is used for mining before... you don't get that much dust and crap inside a normal gamer's pc, although a few did come close in this thread: The Filthy, Rotten, Nasty, Helpdesk-Nightmare picture clubhouse. Some of those pictures always give me a good laugh. However, reflowing might actually improve the situation somehow, as you mentioned tightening the backplate screws actually made the artifacting issue slightly better(although you shouldn't really do that. It might crack the die). I would STRONGLY suggest NOT using an oven though. It heats up the whole card and those resistors on the back might fall off. In addition to that you get a smell of burning crap inside your oven and a risk of melting the 8-pin pcie power connector. You should buy a heat gun and maybe watch some Youtube videos to get a better understanding of how to reflow stuff. Also getting some thermal pads and applying them to the VRAM might help, as artifacts are usually associated with them.

The card works now, I won't touch it. It 100% was owned by a miner, the guy I bought it from told me the story of the card. I don't believe it was used to play games.

"I bought it from a guy, he gave me a week to test it, but i didn't. When I tried to play games with it it crashed. Then I messaged him, he told me that the card didn't have an issue when he used it."

I then asked him, have you tried fixing it? And he told me that he tried underclocking it and it didn't work.

 

[Leooooo1061]

It 100% was owned by a miner, the guy I bought it from told me the story of the card.

I knew it the moment I saw the description. Those mining cards are pretty much always pushed to their limits, it's a miracle this poor card still works at all.
Also, about the thermal pads, this site seems to have a nice list of recommendations Best Thermal Pad for CPU, GPU, RAM, VRM, Laptop and Chipset in 2025. I'm not exactly sure on what pads to buy either; I'm still finding ones for my RX590.

 

[CheteriagalenTriagalnik]

If it works, it works! Think of it from a positive perspective - you're getting lower power draw. Screwing around with it more could lead to something breaking, so just enjoy what you have.

 

[AsRock]

Yeah, I've been watching northwestrepair, he cuts off soda cans and makes a shield with that. Probably gonna do that if the card shits itself again.

I have calipers, I'll measure the thickness. Any ideas what pads to get, what brand?




The card works now, I won't touch it. It 100% was owned by a miner, the guy I bought it from told me the story of the card. I don't believe it was used to play games.

"I bought it from a guy, he gave me a week to test it, but i didn't. When I tried to play games with it it crashed. Then I messaged him, he told me that the card didn't have an issue when he used it."

I then asked him, have you tried fixing it? And he told me that he tried underclocking it and it didn't work.

Just ask the company who made the card.

 

[koralski]

If it works, it works! Think of it from a positive perspective - you're getting lower power draw. Screwing around with it more could lead to something breaking, so just enjoy what you have.

I agree completely, it tested right in the smack middle on Steel Nomad and Time Spy, so i couldn't care less.

P.S : Pozdravi ot BG

 

[raduque]

1080s are still good cards. I'd kill to find one for $5 and repaste it to get a working card, lol.

BTW, this story reminds me of my R9 280x. Bought it off my boss for between $70-85 because it was VRAM artifacting. I repasted it, put heatsinks on the VRAM chips and then eventually flashed a 1.5v VRAM bios (from here, in fact!). It's still going great to this day.

 

[RJARRRPCGP]

Squares=often unstable core.
Same with "display driver stopped responding and recovered" message from Windows.

I am starting to suspect its a bad connection to something. If the card warms up it doesn't crash. But if I take it out of the machine for a day and try and start a game it'll crash.

Exactly what happened the day after I posted the thread, woke up. Installed the GPU in my GF's machine. Ran Cyberpunk, and it crashed.

Sounds like the core needs to be reballed.

 

[CheteriagalenTriagalnik]

1080s are still good cards. I'd kill to find one for $5 and repaste it to get a working card, lol.

BTW, this story reminds me of my R9 280x. Bought it off my boss for between $70-85 because it was VRAM artifacting. I repasted it, put heatsinks on the VRAM chips and then eventually flashed a 1.5v VRAM bios (from here, in fact!). It's still going great to this day.

I recently returned my 5070 Ti, since it nearly set my house on fire and installed a 1080 while I choose something else. Tbh, it still works great and I'm surprised with the amount of things that run flawlessly on it.

I agree completely, it tested right in the smack middle on Steel Nomad and Time Spy, so i couldn't care less.

P.S : Pozdravi ot BG

Pozdravi, bratle!

 

[raduque]

I recently returned my 5070 Ti, since it nearly set my house on fire

Oh dang, that's no bueno. I was considering a 5070 12gb as an upgrade to my 2080.

My 1080 is actually doing LLM and Plex duty in my server.