I'm at a loss here. Out of the blue, my 1080ti kingpin started crashing under anything that pushes it to 100% load for a few seconds.

I've had the card for almost 4 months now and haven't had a problem, with it holding decent overclocks on air. Now it crashes even at stock settings under a minute in Heaven.

I reinstalled latest drivers with DDU, reinstalled PXOC, cleared CMOS, reseated card. More than once.

I was about to try out a different PSU, when I got gpu artifacts and crashed in Widows desktop, so it's clearly not power related.

Underclocking allowed me to finish Heaven, but then it crashed when I Alt-Tabbed to the desktop and I got a blue screen.

The only thing I've done to the card is that I removed the cooler to change thermal paste with Kryonaut, a while before the problems started, so I don't think it's related.. That's all. I'm fairly certain I didn't mess anything up, I've done this before and the process is very simple.

I'm thinking of RMAing, but I'll take any other suggestions before I do that.

Thanks in advance!

Thanks for replying.

I just tested an old Gigabyte 780 Windforce OC I had laying around. Same settings on Heaven.

It completed it without any problems whatsoever, even if it went up to 80C with the default fan curve. So that rules out the PCI-E slot, PSU etc and isolates the graphics card.

If anyone knows, does EVGA still issue Kingpin replacements? It's not listed on the website anymore.

Edit: also completed Superposition benchmark at 4K (8K required 6gb vram, figured it'd crash with the 3gb of the 780), without any hiccups. Used to crash after a few seconds with the Kingpin.

Placed the kingpin back and I'm not even getting picture now, even though the motherboard shows AO after boot.

My first one also started to artifact after 4 months. RMA'd it, and got a brand new in retail box one back from evga. 
 
Note: I've had my replacement for 7 months now. Still working without any issues. 
 
 

HarkerN
You got a brand new after 4 months? Aren't you supposed to receive a repaired one after 3 months?

Yep. Yes, you're suppose to get a repaired one after the first 30 days, but sometimes you can get a new one if they don't have any repaired ones. 

Quick update on the situation. I spoke with tech support and I got an RMA approved. Customer support was really good and the replacement was sent on the same day they received my old card. Shipped it on Monday, it was at EVGA EU on Tuesday morning and EVGA EU shipped my card later that day and I received it Thursday morning.
 
Bonus, I got a brand new card, apparently, since it has the -KR suffix!
 
Unfortunately, the card was practically DOA, with artifacts and driver crashing at stock settings. The GPU was hitting the thermal limit and heating up the memory chips as well. What I would call abysmal thermal performance, with the GPU hitting 76C on stock speeds (before crashing out of literally anything).
 
I considered just sending it back, but I removed the cooler and noticed that the thermal paste was badly applied and a little dried up. I changed it to Thermal Grizzly Kryonaut, put the cooler back on and issue solved! So now the card worked as intended. I get that QA can sometimes overlook something, but the KPEs are supposed to be tested before they go out the door. This is a 1300$ card. But it's working, so no biggie. Just a little annoyed, a lot of people can't do what I did or don't want to (and shouldn't have to).
 
Now to the overclocking part. My previous card would be completely stable with +85 core, +580 memory and go as high as 2088MHz, settling at 2050MHz under sustained load, with memory at 12170MHz (6085).
The new one can't sustain anything higher than 2037 on Heaven, lest it crashes, but overclocks well on the memory, going up to +775 for 12554MHz (6277)!!
 
The strange thing is that the core clock remains unchanged in Heaven as the card heats up. Starts at 2037 and stays at 2037, even as I hit 65C on the GPU. At 2050, I crash, but without hitting any limit, thermal, voltage or power. Any ideas as to why?

Sajin

HarkerN
The strange thing is that the core clock remains unchanged in Heaven as the card heats up. Starts at 2037 and stays at 2037, even as I hit 65C on the GPU. At 2050, I crash, but without hitting any limit, thermal, voltage or power. Any ideas as to why?

Odd, but good. It should definitely go down when 60c is hit. You can most likely get more out of your card by following my oc guide listed .

Yeah, it's quite odd. In certain programs like Heaven and FurMark, it stays at 2037MHz from beginning to end and instead, the power draw slowly increases. In other cases though, as I discovered later, it does drop and it Precision XOC says it hits the power limit, but I set it to 130% and it only goes up to 118% or something like that. I've seen this as being a former bug somewhere, but I didn't have that problem with my previous 1080 Ti Kingpin.

Also, the clock dropped as low as 1949-1962MHz after maybe 10 minutes in a demanding game, with loads around 93%-99% and Prefer Maximum Performance set in Nvidia CP. Isn't the Kingpin supposed to be able to hold at least 2025MHz on air, after being overclocked? I'm running an open case with dual 140mm Noctuas at the front, so airflow and warm air aren't an issue.

I'm going to try your guide either later today or tomorrow. It looks very promising, thanks for taking the time to make it!

I am a little reluctant of flashing another BIOS immediately after receiving the RMA card. Why would I need to do thar just to get the advertised clock speeds? I didn't have this problem before, with my previous Kingpin 🤔.

I haven't had time to try out your guide yet, but I have a funny feeling about this card. Starting with the fact that I had to replace the thermal paste just for it to run out of the box and it's still running hotter then my previous card, reaching 74C in games, in an open case, something I hadn't ever seen before. 😞

What are your temps/clocks, if you don't mind me asking?

Ok, that's reasonably close to my card. I'm hitting 65C overclocked on both core and memory and iirc, you have an open test bench, so that'll make up for the difference.

I don't have a custom loop. I got an NZXT G12 + Kraken X62 to watercool it, but I ran into a stripped screw, one of the ones that hold the midplate in place (one of the four legs).

I'm just disappointed at this point, the card doesn't work as intended on air and I can't watercool it. I called EVGA support and they said that my best option is to cut the corresponding leg off, but it'll void my warranty. Alternatively, I can try to drill the head off the screw, because I realized that the PCB holes are larger than the screws. Hopefully I'll be able to get the midplate out.

But considering the card doesn't work as intended in the first place, I am considering sending it back.

Lol, I'm just showing the "legs". Hey, I'm not pushing the card to LN2 levels, not at all. I'm using the OC BIOS that includes thermal protection. And all sensors on the card read lower temps than with the iCX cooler.
It's a good alternative for someone like me who doesn't have a custom loop.