1080ti stucked 139mhz and 405mhz

[alesiomaso]

I tell you that I bought a second-hand gpu that seems to be faulty, I have been looking for thousands of solutions in this and other forums, I have seen that many people have run into this problem but none find the solution. The problem is that the core clock remains capped at 139mhz and the memory clock does not exceed 810mhz and the performance in games and beanchmarks are really bad, 12 FPS in Furmark
I have already done different tests such as putting high performance mode in Windows and in the Nvidia panel, using DDU, reinstalling drivers, a clean installation of windows, using MSI Afterburner, Asus GPUTweekII, after trying various things I was forced to try flashing a BIOS taken from Techpowerup, but it didn't solve anything.

So far the only thing I have found is that in GPU-Z the Power Draw Board is +400w and PCIe Slot Power +390w DOING NOTHING, simply on the desktop, a Perfcap reason: Pwr, and a continuous green line. It also strikes me that the voltage is constantly 0.65v (I don't know what the normal voltage is).

I am currently using an Aerocool Cylon 600w 80+ bronze PSU, I know it is a bit fair but the previous owner told me that he tested it with an 850w 80+ source, so we could rule out a power issue from the PSU, already that in neither of the two systems has worked well.

I hope you can help me, and thank you very much in advance !!

AMD Ryzen 5 2600x
Asrock b450m Steel legend
2x8GB RAM Corsair vengeance

 

[1986nath]

Try it with a good psu if you can, aerocool are made with cheap chinese parts.

 

[alesiomaso]

Try it with a good psu if you can, aerocool are made with cheap chinese parts.

But as I have commented, the gpu was also tested with an 850w one, also the aerocool PSU would have to be more than enough to move the graphics card. Looking at the photos I think the problem goes beyond the PSU, drivers or temperatures

 

[Hemmingstamp]

But as I have commented, the gpu was also tested with an 850w one, also the aerocool PSU would have to be more than enough to move the graphics card. Looking at the photos I think the problem goes beyond the PSU, drivers or temperatures

Tried the card in a friends machine?

 

[Hachi_Roku256563]

was it like shunt modded?

 

[alesiomaso]

was it like shunt modded?

I don´t believe it

 

[GerKNG]

this card has a problem with it's power delivery or the shunts.
the card is stuck in a safe mode because it has no idea what is going on.

 

[Falkentyne]

Try it with a good psu if you can, aerocool are made with cheap chinese parts.

This has absolutely nothing to do with the PSU. Please don't make people throw away money for your own guessing and experiments. Not everyone on these forums is filthy rich.

The card is damaged or defective. Either a shunt mod that went wrong, or something otherwise physically or electrically damaged on the card. The 8 pin power connectors seem to be reporting properly at least.
Might want to very carefully examine the card for any evidence of tampering or modding, replaced shunt resistors, solder bridges shorting something, anything of that sort.

There was a user with a brand new retail 3080 who had the exact same problem. Card was simply defective out of the box and had to be RMA'd.

 

[Final_Fighter]

can you provide a picture of the card?

 

[alesiomaso]

This has absolutely nothing to do with the PSU. Please don't make people throw away money for your own guessing and experiments. Not everyone on these forums is filthy rich.

The card is damaged or defective. Either a shunt mod that went wrong, or something otherwise physically or electrically damaged on the card. The 8 pin power connectors seem to be reporting properly at least.
Might want to very carefully examine the card for any evidence of tampering or modding, replaced shunt resistors, solder bridges shorting something, anything of that sort.

There was a user with a brand new retail 3080 who had the exact same problem. Card was simply defective out of the box and had to be RMA'd.

Thanks for understand!! I have examined the gpu and it seems that everything is fine, there are no signs of having been modified or attempts to shunt mod and there appear to be no damaged capacitors. I have noticed a very slight but VERY SLIGHT "corrosion", although I don´t know if it is corrosion, they are slight stains, could this be the problem? As I told you, I do not know if it reaches the point of classifying corrosion because they are almost invisible.

As you know I cannot have RMA

can you provide a picture of the card?

 

[80251]

To see if the card was improperly shunt-modded it would be necessary to take it apart -- at least partially. Do you trust the previous owner?

 

[alesiomaso]

To see if the card was improperly shunt-modded it would be necessary to take it apart -- at least partially. Do you trust the previous owner?

I don't trust him haha.
I completely disassembled it and I don't see anything strange

 

[agent_x007]

GPU-z clearly indicates you pull 400W out of PCI-e slot (which is impossible, hence card's turbo got confused).
Because of this, I think shunt on your PCI-e port may be dead or damaged (replacement may be needed).

 

[Gungar]

I have the same problem with the same card (Asus strix 1080ti) it's a hardware problem but i can't find the source of it, i am not an expert. I am looking for a professional to examine the board but it's hard to find someone with those skills in Switzerland.

I have a corsair AXi1500, i can garantee you that's not a psu issue.

this card has a problem with it's power delivery or the shunts.
the card is stuck in a safe mode because it has no idea what is going on.

That's also what i think it is.

 

[pavle]

If nothing else, Asus was known to not put any heatsinks on RAM of their Kepler, Maxwell and Pascal cards. And RAM is being fed power through PCIe connector. So something must have gone haywire there...
Is there any discoloration or brownness around any power supply components (slightly burned PCB)?

 

[mav2000]

I had the same issue with a 1070ti from MSI, a few months ago. I sent it in for RMA/repairs and they told me the card was dead. I would suggest you RMA it if possible.

 

[FireFox]

I would suggest you RMA it if possible.

This:

As you know I cannot have RMA

 

[alesiomaso]

If nothing else, Asus was known to not put any heatsinks on RAM of their Kepler, Maxwell and Pascal cards. And RAM is being fed power through PCIe connector. So something must have gone haywire there...
Is there any discoloration or brownness around any power supply components (slightly burned PCB)?

No sign of anything like that

I have the same problem with the same card (Asus strix 1080ti) it's a hardware problem but i can't find the source of it, i am not an expert. I am looking for a professional to examine the board but it's hard to find someone with those skills in Switzerland.

I have a corsair AXi1500, i can garantee you that's not a psu issue.



That's also what i think it is.

I think that maybe with an volimeter we can find the problem and replace

I had the same issue with a 1070ti from MSI, a few months ago. I sent it in for RMA/repairs and they told me the card was dead. I would suggest you RMA it if possible.

Surely the card have a slighly fault that can be repaired, apparently the card is fine, but there must be a small flaw in the pci connectors

 

[mav2000]

No sign of anything like that


I think that maybe with an volimeter we can find the problem and replace


Surely the card have a slighly fault that can be repaired, apparently the card is fine, but there must be a small flaw in the pci connectors

Since my card had a few days of warranty left, I didnt want to open the heatsink and do any work myself. Sinceww you dont have any warranty left, you can remove the heatsink to see if there is anything wrong, any signs of burn etc.

 

[FireFox]

you can remove the heatsink to see if there is anything wrong, any signs of burn etc.

I see you are missing some post

Here

I completely disassembled it and I don't see anything strange

 

[ERazer]

am i missing something here? theirs no GPU load run a benchmark and check the clock

 

[alesiomaso]

am i missing something here? theirs no GPU load run a benchmark and check the clock

As I said in the post, I have already tried games and beanchmarks, the clocks are stuck at that frequency and it has a terrible performance.
GPU-Z marks a consumption of 400w and Furmark also, there is the problem

 

[Falkentyne]

As I said in the post, I have already tried games and beanchmarks, the clocks are stuck at that frequency and it has a terrible performance.
GPU-Z marks a consumption of 400w and Furmark also, there is the problem

Can you post clear, high resolution pictures of the PCB of the disassembled card, both front and backside? Especially the areas around the shunt resistors.

 

[alesiomaso]

Can you post clear, high resolution pictures of the PCB of the disassembled card, both front and backside? Especially the areas around the shunt resistors.

 

[oldwalltree]

PCB looks fine. No signs of modding from what i can tell?