• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

GPU Failing, Looking for a good BIOS update?

@Laezur I have prepared you low power BIOS for your card and remembered the GTX 980 Ti has this rather evil VRM design where the middle elements are smaller than the outer and they get cooked the way caring nvidia routed the power through it (they are expected to carry the same current as the more powerful ones), meaning neither the GPU is bad nor the RAM, but VRM is tired.
I believe this BIOS should help or you'll need an expert to solder you some more powerful elements into VRM is all.
I hope this BIOS helps though (more sensible power and temperature targets and it will start with more normal clock for such a big chip; I use much lower % with my GTX 680 and 980 yet they work as factory so there's still room).
 

Attachments

@Laezur I have prepared you low power BIOS for your card and remembered the GTX 980 Ti has this rather evil VRM design where the middle elements are smaller than the outer and they get cooked the way caring nvidia routed the power through it (they are expected to carry the same current as the more powerful ones), meaning neither the GPU is bad nor the RAM, but VRM is tired.
I believe this BIOS should help or you'll need an expert to solder you some more powerful elements into VRM is all.
I hope this BIOS helps though (more sensible power and temperature targets and it will start with more normal clock for such a big chip; I use much lower % with my GTX 680 and 980 yet they work as factory so there's still room).
Hi, thank you so much for this!

Unfortunately I am receiving these errors when trying to flash this ROM.

See attached.
error 1.png
error 2.png
 
Of course, you need the nvflash with certificates bypassed: HERE.
You disable the GPU in the Device Manager, then open Admin Command prompt (right click Open with Admin Rights) and nvflash biosname.rom.
 
Of course, you need the nvflash with certificates bypassed: HERE.
You disable the GPU in the Device Manager, then open Admin Command prompt (right click Open with Admin Rights) and nvflash biosname.rom.
Thank you!
 
I wouldn’t say I’m disalussioned, I’d say I’m lost, and desperate to find a fix that doesn’t include spending money lol. Was willing to try last resorts. I don’t HAVE money to spend, my circumstances don’t allow, and my PC is a big part of my life unfortunately. Without going into detail.

Anyway, I’ve tried it now, it didn’t work as expected LOL.
Unfortunately some things just break from usage.

That said, video cards don't break all that fast.
They break very slowly if they're kept in dust free environments like a case with proper dust filters & airflow setup.

You don't need to do anything exotic with PC hardware contrary to popular internet belief and in most cases it will be counterproductive. Somehow the first go-to for people who experience worse GPU performance is that they update BIOS and repaste their cooler, voiding warranty left or right. It's never been a fantastic idea, in general.

I'm running a 1080 since 2017, all I ever did was blow it out once a year. Most years, all that comes off is a thin layer of dust, nothing caked, just thin dust. Of course ambient conditions may vary, but it all starts with decent air quality in your room and a dust filtered case. Beyond that its just luck of the draw, plus the strain you've put the hardware under. If Memory has been operating close to temp limits for a long time or has the chance of exceeding it momentarily during normal use, you're going to find cards lasting far shorter. Here, again, dust comes into play as it can push cards out of their safe tolerances, perhaps not on the GPU die, but elsewhere on the pcb. There are notable examples of memory dying for lack of cooling. EVGA especially had some cool (or hot... :D) stories to tell there :)

If you want GPUs to last, when you see that cooler performance slowly inch back, or temps slowly creep up, its time for the simple cleaning job. If you keep losing a few mhz on your OC to remain stable, its time to dial in a slightly lower power target / clock down a bit. The 1080 I run now does it at 90% power target; the net loss of performance is 5% at best; temps are still royally below 80C.

You could also repaste it at that point - but that's introducing risk, for what is essentially marging of error performance loss.

TL DR; try those lower power targets, remove your OC, and see where you can get stability again. Also, consider the options of selling off whatever life is left in it. An honest ad with an honest price might fetch you more than you expect today. Even if just for a collector ;)

And if money is short, consider second hand markets for your next card. It might end up feasible to get a Pascal card or even an x60~x70 Turing as people start upgrading now. They'll be good for another few more years for most things, especially at 1080p~1440p they destroy most content still, today.
 
Last edited:
Of course, you need the nvflash with certificates bypassed: HERE.
You disable the GPU in the Device Manager, then open Admin Command prompt (right click Open with Admin Rights) and nvflash biosname.rom.
Thank you so much for your effort and time making the BIOS for me. I succefully flashed the card, but unfortunately I still lose signal during the benchmark so I think it is safe to say that it is just time for it to be put to rest and a new one be looked at.
 
Thank you so much for your effort and time making the BIOS for me. I succefully flashed the card, but unfortunately I still lose signal during the benchmark so I think it is safe to say that it is just time for it to be put to rest and a new one be looked at.
Did you try with another PSU?
 
Did you try with another PSU?
Yeah that’s the first thing I tried, same issue. The second PSU I tried was brand new unused but unfortunately it’s a poor brand, some Japanese brand called Kronus. Only 80 rated Bronze 700W so the predicament I’m in is that I still don’t for 100% sure know it’s NOT the PSU but I don’t have any more money to try these things
 
Yeah that’s the first thing I tried, same issue. The second PSU I tried was brand new unused but unfortunately it’s a poor brand, some Japanese brand called Kronus. Only 80 rated Bronze 700W so the predicament I’m in is that I still don’t for 100% sure know it’s NOT the PSU but I don’t have any more money to try these things
Can you use MSI afterburner and lower your vRAM speed , lets say to half, and see if it goes better? Same with the GPU?
You should try to reset your motherboard BIOS settings to default, if nothing helps your GPU is probably dying.
You could also ask a friend and test your GPU in his system to be sure its 100% the GPU fault.

Edit: If nothing helped , something that i dont recommend but ive heard some other ppl had success with ,is to bake your GPU at some temperature , it can "resolder" itself or something (thats something to do as last resort before trashing the GPU)
 
Edit: If nothing helped , something that i dont recommend but ive heard some other ppl had success with ,is to bake your GPU at some temperature , it can "resolder" itself or something (thats something to do as last resort before trashing the GPU)
I would add that baking could put short circuits on the card which could damage the motherboard. I wouldn't try it.
 
I would add that baking could put short circuits on the card which could damage the motherboard. I wouldn't try it.

If that happens , even your system may be at risk of damage then.

Also personally I would never try the "baking" method, then I just accept it's the end and I need to look for a new GPU.
 
Can you use MSI afterburner and lower your vRAM speed , lets say to half, and see if it goes better? Same with the GPU?
You should try to reset your motherboard BIOS settings to default, if nothing helps your GPU is probably dying.
You could also ask a friend and test your GPU in his system to be sure its 100% the GPU fault.

Edit: If nothing helped , something that i dont recommend but ive heard some other ppl had success with ,is to bake your GPU at some temperature , it can "resolder" itself or something (thats something to do as last resort before trashing the GPU)
Holy shit I actually just made it through the benchmark for the first time..

I have no idea, how, or if what I applied in afterburner actually helped but I underclocked as much as I could and I reduce the power limit.

That's the first time it's managed to get through the entire Heaven Benchmark since I started running into the problem...

afterburner.png
 
Holy shit I actually just made it through the benchmark for the first time..

I have no idea, how, or if what I applied in afterburner actually helped but I underclocked as much as I could and I reduce the power limit.

That's the first time it's managed to get through the entire Heaven Benchmark since I started running into the problem...

View attachment 274343
Try to increse the GPU clock first and see if it crashes if not rise your vRAM speed till it crashes, and go with some stable underclock, i think your vRAM is starting to fail.

I would add that baking could put short circuits on the card which could damage the motherboard. I wouldn't try it.
If you google it you will find that this fixed the problem for some people , i think Linus from LTT did something similar in some video but as i was saying i wouldnt recommend it and he can do it at his own risk , some people really dont have any other option so i thought it worth mentioning.

Edit: I think his vram is going bad if it works when downclocking so probably this wont do anything for him anyway, i dont know.
 
Last edited:
Try to increse the GPU clock first and see if it crashes if not rise your vRAM speed till it crashes, and go with some stable underclock, i think your vRAM is starting to fail.


If you google it you will find that this fixed the problem for some people , i think Linus from LTT did something similar in some video but as i was saying i wouldnt recommend it and he can do it at his own risk , some people really dont have any other option so i thought it worth mentioning.

Edit: I think his vram is going bad if it works when downclocking so probably this wont do anything for him anyway, i dont know.
Is it worth even raising my clock at all from where I’ve managed to get it to run the benchmark? I literally just want to play world of Warcraft, and if it will at this setting I’d be happy to leave it until I can afford to buy another one lol
 
Is it worth even raising my clock at all from where I’ve managed to get it to run the benchmark? I literally just want to play world of Warcraft, and if it will at this setting I’d be happy to leave it until I can afford to buy another one lol
I think that its only the vRam that needs underclocking , you can just go back to your stable settings if it fails but its your call if you dont mind to leave some performance on the table.
 
Interesting; is it now stable and underclocked in addition to my LP BIOS?
Just keep testing it and use it if it works.
I can later lower the max boost clocks in BIOS, I don't think memory is the problem, just the core - it requires much more from VRMs. We shall see. :toast:
 
Interesting; is it now stable and underclocked in addition to my LP BIOS?
Just keep testing it and use it if it works.
I can later lower the max boost clocks in BIOS, I don't think memory is the problem, just the core - it requires much more from VRMs. We shall see. :toast:
So with your BIOS and these afterburner settings I'm able to consistently complete the benchmark.


I also got the most game time I've been able to get out of WoW, at about 10-15 minutes before no signal (still having sound until I can hear the system freeze in the background behind the black screen).

I'm confused as to why I can run the benchmark, but playing WoW won't work.

afterburner.png


image0.jpeg
 
My guess is a cracked solder joint, so it depends on thermal expansion; the benchmark keeps things hot, the game has the temperature vary.
 
Last edited:
Is it worth even raising my clock at all from where I’ve managed to get it to run the benchmark? I literally just want to play world of Warcraft, and if it will at this setting I’d be happy to leave it until I can afford to buy another one lol
The gpu is failing, replace it. RX 6600 or newer
 
Had to do this with an old GTX 560 SE. Lowering both the GPU and Memory clocks helped fix this issue for good. (At least until it goes bang.)
 
Despite what everyone was saying, I did the flash anyway as a peace of mind thing. Obviously didn't work, lol.

I guess it is time for an upgrade... I can't afford one right now, but I'll save some money .

Annoying because I only want to play World of Warcraft, and I'm itching to play it right now not in however many months.
Mind you, people telling you to not flash the BIOS were genuinely trying to help. Just do search on these forums for threads asking to recover from bad GPU BIOS flashes. Plus, it's not like the GPU BIOS breaks down and a refresh will make it work properly again.

Edit: And I agree, it's still a terrible time to be shopping for a video card :(
 
So with your BIOS and these afterburner settings I'm able to consistently complete the benchmark.


I also got the most game time I've been able to get out of WoW, at about 10-15 minutes before no signal (still having sound until I can hear the system freeze in the background behind the black screen).

I'm confused as to why I can run the benchmark, but playing WoW won't work.

View attachment 274349

View attachment 274350
Your GPU speed is low but your vRAM is still HIGH , you need an even lower memory clock to test.
 
Back
Top