I may have broken my 1080 Ti

Are these measurements taken while it's idling, or actually doing something? Try taking another measurement while it's doing something that would normally cause it to crash. Edit: It's really the 12V that we're interested in.

The six and eight pin connectors only provide 12V, but the system also provides 3.3V to the card through the slot. This is frequently used to power things like the BIOS chips and VRM control logic. On this card, it may also be used to power the transistor gates, and as I mentioned before, if the gate voltage is low, the transistors will run hot. If you look at the PCI-E pinout, you'll see that pin 8 on both sides of the connector is the +3.3V input. You'll need to follow the traces to somewhere you can probe with the card plugged in.

This may be a dumb question, but... have you tried this card in a different system, or a different card in the system in question?

Ok the 3.3V will be harder to check but I'll give it a try. Under load it drops to 11.95V, which is in reading variance to the 12V at idle I was measuring earlier.

I tested it in another system, and let me tell you intermittent problems are the worsts! It ran OCCT for 30 minutes, no problem, then Furmark (I know) for another 30 minutes without problem. Then I went back home, plugged the card in my system, played DOOM for an hour without problem. So the result of testing it in another PC is inconclusive. I will need to test it for a longer period of time but it was all I could do because of the curfew here.
I sadly don't have another card to put in my system.

As I see it, I need more time and hardware to test everything, cause all I did today didn't help.
 
Ok the 3.3V will be harder to check but I'll give it a try. Under load it drops to 11.95V, which is in reading variance to the 12V at idle I was measuring earlier.

I tested it in another system, and let me tell you intermittent problems are the worsts! It ran OCCT for 30 minutes, no problem, then Furmark (I know) for another 30 minutes without problem. Then I went back home, plugged the card in my system, played DOOM for an hour without problem. So the result of testing it in another PC is inconclusive. I will need to test it for a longer period of time but it was all I could do because of the curfew here.
I sadly don't have another card to put in my system.

As I see it, I need more time and hardware to test everything, cause all I did today didn't help.
Have you considered that maybe your reinstallation of the heatsinks actually isn't proper? Like, I know you said you're confident that it is, but is this the first time you've actually done this on a graphics card? Can you pull the heatsink off and post of photo of the top of the card and bottom of the heatsink with the thermal pads and grease still on there?

That 11.95V is actually kind of troubling. Slayer the TX650 droops down to around 11.89 under load, as I recall. I can't say for sure that that power supply is the root cause of your problems, but it certainly isn't helping. I know now isn't the best time to be in the market for hardware, but you should consider replacing it at your next upgrade, if not sooner.
 
I know now isn't the best time to be in the market for hardware, but you should consider replacing it at your next upgrade, if not sooner.
Buying a new PSU isn't much of a problem. I was even debating it a couple weeks ago cause it's all what is missing to have another complete system I could use for tests.

It's my first time on a GPU. I'm confident it's ok but that doesn't mean much. I'll pull it off and take pictures tomorrow.

EDIT: Forgot to say, the 8 pin connector drop to 11.95 - 11.97v, while the 6 pins drop to 11.99v
EDIT2: Running OCCT Power test I get 11.93 - 11.94v at the 8 pins and 11.95 at the 6 pins. I could swap cable to see if the bigger drop follow the cable, or the connector on the card. Another thing to test tomorrow. I attached the values seen by OCCT
 

Attachments

  • OCCT.png
    OCCT.png
    25.3 KB · Views: 0
Last edited:
So another day of testing and I think I solved it! RazorWind you were right about the PSU. I tested the card in another system, 12.06v at idle, 11.99 under load.
I ordered a Corsair RMx 750w today.

So long story of what happened:
About 2 years ago, I realized the fan on my PSU stopped working. I hacked it by soldering a fan header on the wires, and plugged it on the motherboard. I added a resistor to the wires to have a correct fan speed.
In February I retired my i5-4670K and upgraded to an AMD 3600, while I was waiting for my 5600X. I kept the same PSU, but the new motherboard allows me to control fan speed, so I played with the PSU fan speed and maybe set it too low. I was using the system temperature, and with everything water-cooled, this temperature was not going very high.
The PSU over heating may have been the reason for my crash when running 3DMark after water-cooling the card. It may explains why even with my old 7800GT I was not able to boot, for a couple hours, while the PSU was cooling down. After a couple hours, everything worked fine.
Then 2 days after I received my 5600X and installed it. It did reset my BIOS settings and also fan speed. I set them back to about where they were, but it seems 0.1v on my PSU fan can make a huge difference. I bumped up the PSU fan speed today and now my system is stable. I'll wait a couple days to confirm.
Except I see my 12V rails may have took a hit, or maybe they are just tired. Anyway, this PSU is going to be used in a test bench when I get the new one next week.

As I say since day one, I can't have broke the card doing what I did, and that was right, it wasn't the card, it was the PSU. Now will I try to water-cool it again, I'm not sure, especially since I need a working computer for work, and the price of those cards right now is way to high in case something bad happens again.

So I want to thank everyone who helped, especially you RazorWind. You've been really supportive and gave a lots of informations that pointed out what the culprit really was.
Also for the side thread about washing parts, I did wash my 7800GT in the sink, with dish soap, cleaner than new and it still works.
 
Right on! Depending on what you're planning on testing with that test bench, you may want to rethink using that power supply, though.

Once you're sure that the new power supply fixed the problem, I'd absolutlely recommend water cooling this card, but I'd probably also try to convince you to use a full coverage water jacket. It may be hard to find a new one for a 10 series card now, but this board should be the same as a Founder's Edition, so it shouldn't be that hard. I've got one for a 1080 (non-Ti) lying around that I'd just give you, but I don't think it would fit your card, because the Ti is different.
 
I don't know what will be tested either. I'm the one fixing every friends and family PC so I'm sure it will be useful at some point.

In my first post I linked an article. At the top of the article it is written "To make matters worse, I learned that the PCB was not the 1080 Ti reference PCB. It was very similar, but just slightly off. See below." Then follows a picture of the reference board and my PCB.
Bykski have a full coverage block => https://www.bykski.us/collections/g...rage-gpu-water-block-clear-n-as1080ti-turbo-x
Supposedly it's compatible with non-Ti, but is yours for an Asus Turbo?

I got a crash tonight while casting a movie. Maybe it's not only the PSU temperature, but that the 12V rails are dying.
 
So I got the new PSU today. Curiously enough, I get the same voltage at the card than with my previous PSU, and sadly, I got a crash again while casting a movie.
At this point I'm thinking it may be my 5600x, I will try to put the 3600 back on since I had no problem with this CPU.
 
So I got the new PSU today. Curiously enough, I get the same voltage at the card than with my previous PSU, and sadly, I got a crash again while casting a movie.
At this point I'm thinking it may be my 5600x, I will try to put the 3600 back on since I had no problem with this CPU.
Possible, but unlikely. What happens if you try to play games or run a benchmark now?
 
It's a thing, but maybe the thread should stop being derailed
I think it deserves it's own thread, but it also keep mine active so I don't mind

Possible, but unlikely. What happens if you try to play games or run a benchmark now?
I had no time to play game, but I did run TimeSpy Extreme, scored 100 points above the average of the same combo CPU / GPU. Maybe it would have crashed if I had let it run for multiple pass. I'll try to play some game tonight and see what happens.
I removed the overclock on the CPU this morning (66MHz on the memory controller), even disabled DOCP, PBO, etc. We'll see. It crashed very often while video conferencing at work (and that's a big problem), or while watching a movie.
 
I played DOOM for over an hour tonight, no issue at all. But I found an interesting glitch. I loaded a 4K video on YouTube, and another one in VLC to work the Video Engine.
Since I have this card I'm using NVIDIA Inspector to downclock the card when idling. If I don't have this program running, the GPU clock will stay at base clock instead of falling down to 600MHz.

If I check the "Activate Video by VPU Usage" option, I loose all my screens. I recorded a video here


I click way to hard on the mouse so you can hear the click, and a fraction of a second later I loose or recover the screens.

I'm not sure if it's something else, but so far most of the time I had crash were while playing a video, or a video conference at work, all things that work the video engine and not the 3D Engine. Not sure if I'm on to something.
Also you can see the voltages reported by HWInfo at the end. They are the same than with a DMM. That's with the new RMx PSU. Curiously enough, when I tried it in another system, they never dipped under 11.98v
 

Attachments

  • Untitled.png
    Untitled.png
    41.5 KB · Views: 0
Personally I wouldn't worry about that voltage drop if that's the minimum reading. This is from my 1300w Cooler Master V3 Platinum powering my 3090 which pulls 530w. Depending on how the card reads the voltage and where it's reading it from it won't be entirely accurate. Those lows are from when it's under load in Time Spy.
1617849517991.png
 
Last edited:
They are from when the card was under load indeed, in TimeSpy, or while gaming. You still get a minimum of 12.0v, while I drop to 11.92v, which was the reason I bought another PSU. When I tested the card in another system it would not drop under 11.99v, tested with a voltmeter and HWInfo.
 
I played DOOM for over an hour tonight, no issue at all. But I found an interesting glitch. I loaded a 4K video on YouTube, and another one in VLC to work the Video Engine.
Since I have this card I'm using NVIDIA Inspector to downclock the card when idling. If I don't have this program running, the GPU clock will stay at base clock instead of falling down to 600MHz.

If I check the "Activate Video by VPU Usage" option, I loose all my screens. I recorded a video here


I click way to hard on the mouse so you can hear the click, and a fraction of a second later I loose or recover the screens.

I'm not sure if it's something else, but so far most of the time I had crash were while playing a video, or a video conference at work, all things that work the video engine and not the 3D Engine. Not sure if I'm on to something.
Also you can see the voltages reported by HWInfo at the end. They are the same than with a DMM. That's with the new RMx PSU. Curiously enough, when I tried it in another system, they never dipped under 11.98v

This might be a dumb question, but have you tried using DDU and then installing the drivers again, or doing a clean Windows install?

Did you get these symptoms when you tested the card in the other system?
 
This might be a dumb question, but have you tried using DDU and then installing the drivers again, or doing a clean Windows install?

Did you get these symptoms when you tested the card in the other system?
I did not. Reason is I run a dual boot: one for personal stuff, and the other is work only. Both of them started crashing after my failed water cooling attempt / CPU upgrade. I'd be very surprised if both Windows install started having the same problem at the same time, after playing with the hardware.

I haven't test the Video Engine glitch in another system as I discovered that yesterday night, but I'll test it on my bench as soon as I can, most likely Friday night.
 
I don't think we're looking at the same picture ... 11.68, 11.69, 11.77 all in the screen shot. Your card is roasted dude.

I'm looking at the 8-pin since I don't have those "Misc" on mine and have no idea what they are for. What make you think it's roasted more precisely?
 
I'm looking at the 8-pin since I don't have those "Misc" on mine and have no idea what they are for. What make you think it's roasted more precisely?
You have fairly consistent crashing behavior that occurs in multiple software environments, which are presumably set up correctly. The card mostly works, so this likely isn't a power issue on the board, you've tested with multiple systems and power supplies, and that basically leaves the GPU die itself as the remaining potential problem source.

You should still try it on as clean a windows install as you can manage, to rule out the possibility that this really is a software problem, but I'm starting to agree with lobstar that the problem here is most likely a GPU failure. :(
 
You have fairly consistent crashing behavior that occurs in multiple software environments, which are presumably set up correctly. The card mostly works, so this likely isn't a power issue on the board, you've tested with multiple systems and power supplies, and that basically leaves the GPU die itself as the remaining potential problem source.

You should still try it on as clean a windows install as you can manage, to rule out the possibility that this really is a software problem, but I'm starting to agree with lobstar that the problem here is most likely a GPU failure. :(
More specifically, Blue has done an excellent job at troubleshooting the issue and has shown the ability to reproduce it in a very exact fashion. I believe you've narrowed the issue down to exactly which part of the device failed. It's possible the games you are playing are displaying some encoded video during cut scenes or as part of the game assets which possibly is triggering your issue haphazardly while gaming. However, the way it fails in that very specific use case Blue showed in the video is what makes me think that a specific part of the card has failed. I bet you could find a use-case where the card would never show those issues and you might get use out of it but combine that with the temps and other ancillary issues that have been described in the thread I don't think there is much chance of saving it long term.

I agree with the idea of a minimal testing environment where you have a fresh install of windows, latest drivers, and minimal extraneous software outside of that which is needed for testing. Focus on reproducing that test showing the screens going blank.

Sorry for you loss. I'd be trying everything I could as well in this current video card market.
 
Intermittent problems are so hard to diagnose. I had no crash in the last two days. It did multiple hours of video call, some gaming, everything is fine, but it doesn't mean it's fixed. What I have done is disable every performance option on my CPU, so no DOCP, no PBO, nothing. I'll give it a few more days, then enable those options one at the time, and let a couple days between them. Maybe the video engine issue is unrelated.
 
So I figured I'd keep you updated on this issue. It was not the video card after all, it was my OC on the CPU (+66 on memory, and maxed-out PBO, which was fine one my 3600) . I tested it after every setting I changed and always passed the test, but the video didn't. I had no crash in about two weeks now. The video engine problem was unrelated after all. It was a tough one to diagnose!

I'm getting a full coverage water jacket today for my 1080 Ti and will start this project next week-end.
Thank for the support everyone
 
Back
Top