Can I fix an artifacting card?

reverendbud · December 5, 2019

Hello, fellers,

Being on somewhat of a budget, I purchased a refurbished 1080 Ti from Zotac (only $330 USD!). Coming from a GTX 1060 6gb, I'm feeling pretty excited.

Was delivered today, so once I'm home from work, I swapped out the GPUs. Boots up fine, threw it on a benchmark (userbenchmark, to be specific) and after the first 2-3 tests, both my monitors had artifacts and my PC froze up.

I reseated the card and the two PCI-E 8-pin power cables to make sure it's in there all nice and good, plus fresh drivers, no GPU OC software installed, but same issue. HWiNFO showed me I was right at about 60C when it artifacted.

I swapped my GTX 1060 back in and everything's all nice and good again - no artifacting or anything.

I'm running a Ryzen 7 3700X, 32GB DDR4 3600mhz, on a Thermaltake Smart 750W PSU. I am slightly concerned that my PSU isn't able to supply enough power, or maybe even the wall socket, but from what I've read, it's more likely to be faulty VRAM or a cooling problem. Plus this PSU supplies 62A @ 12V, and the 3700X is 65W TDP. A quick wattage calculation on Newegg indicates I should be consuming somewhere in the realm of 450-500W. (There actually isn't an option to calculate the power needed for a 1080 Ti on the Newegg wattage calculator so I used the 2080 Ti instead.)

Well, I don't really know why I'm making this post, as I made the RMA request for a replacement earlier today. I am really tempted to open it up and see what state the thermal paste is in and how good the coverage is. I've read some posts where a user found that one small corner of the die wasn't covered by paste and that caused the card to artifact. Also wanting to make sure every thermal pad is looking good and looks like it had been making good contact. The card was PRETTY HOT after using it for like 10-15 minutes but I've never owned a beefy GPU before. There's these "power boost" modules on the backside of the PCB that were particularly hot too. I almost want to prove to myself that I could fix it but if the VRAM is bad... then RIP.

What's your take? Bad VRAM or should I check things out?

Bombastinator · December 5, 2019

If it’s GPU hardware their “refurbish” wasn’t much to write home about. Should be known good which implies onboard problems except refurb quality is unknown. As you say, if it’s half out the door already I guess it’s moot. Those things still going for $330? It’s tempting if they got a dual link dvi port on em.

reverendbud · December 5, 2019

2 hours ago, Bombastinator said:

If it’s GPU hardware their “refurbish” wasn’t much to write home about. Should be known good which implies onboard problems except refurb quality is unknown. As you say, if it’s half out the door already I guess it’s moot. Those things still going for $330? It’s tempting if they got a dual link dvi port on em.

Well, I wasn't impressed, of course. However, my curiosity got the best of me and I opened her up. Replaced the thermal paste on the die with a thin sheet (it was fresh before but a ton was caked on there,) checked out all the thermal pads (they were all brand new,) cleaned up the PCB & VRAM (it was pretty clean besides a few dirty spots, and what looked like small bits of thermal paste on random bits around the card,) and screwed it all back together. Some of screws were loose upon initial tear-down, but not anymore.

So yeah, it's working totally fine. Don't want to count my eggs before they hatch but I'm running on it now. Was hitting just under 85C in Red Dead on all ultra everything 1080P. I hit just under 10k on 3DMark Timespy and hit around 75C, and I didn't even have GeForce drivers installed... Just was testing stability.

They go for $440 normally but there was a Black Friday sale happening and they were $330. I see they've gone down that low before at some times. This one has 3 x DP, 1 x HDMI, 1 x DVI. Using the DVI to power a 144hz 1080p monitor if that helps. LOL

Not sure if something happening during shipment or Zotac's QA but here I am. YMMV.

reverendbud · December 6, 2019

RIP - the card worked fine from the time I put it back together yesterday until about an hour ago from now (24ish hours.) Was playing CS:GO, then I got artifacts and my computer froze. Put on EVGA's OC Scanner, ran a memory test (all graphics card options set to default!), then it froze a few minutes in.

The weird thing here is that the card did work fine after attempting to fix the heat issues... My GPU core temp was hovering around like 60C. As I mentioned, VRAM and VRM temps aren't being monitored here (iCX sounds nice right now.) I'll guess and say the VRAM isn't being properly cooled. If I get the card back and it starts artifacting past the replacement period... I'll get some thermal grizzly pads for these VRAM modules.