Any chance for RMA? What is the rest of your system like and more importantly what power supply are you using.

Memory? Maybe.
I had an XP PC shutting down after a certain RAM size was hit.
Furmark might not need that much of video memory and runs fine.

Hm. I had a Palit GTX 780 in the past that started doing the exact same thing after about one and a half year of use. It still had warranty and I sent it back.

Sounds more like a core or memory issue or defect than anything else. it's very difficult to diagnose something like this and usually the solution would have been RMA. did this card always crash your friends computer? I believe running the benchmarks should rule out PSU, vrm, or overheating issues. Some things I would try, but not certain at all to solve the issue are

Run DDU and test a few other, older drivers. You may want to go back a year or two, just to be sure. Sometimes newer drivers start breaking older cards.

Underclock both the core and memory and see if this cuts out the crashing. If it does it's likely you'll need to increase the voltage to be stable at stock settings.

Also, its been a while since ive owned an nvidia card so im unsure of the procedure, but I would disable low power states, its possible that its crashing in games that don't force the card into its highest clock and voltage settings.

Chances are low but you may find some luck.

Maybe under/overclock and see how it reacts. Or try another bios version or maybe (if it exists) a bios from a version with less memory.

Or if it is possible adjust the voltage by a voltmod or bios mod.

meljor wrote:

Maybe under/overclock and see how it reacts. Or try another bios version or maybe (if it exists) a bios from a version with less memory.

Or if it is possible adjust the voltage by a voltmod or bios mod.

Actually there is a bios revision 84.00.41.00.1C that came out in 10/2015 that i can find. Thats a good idea and worth a shot if he's on the older version.

https://www.techpowerup.com/vgabios/?manufact … odel=GTX+980+Ti

Im doing this from my phone which I'm not too good with, hopefully link works

Undercloking with MSI Afterburner might help. It did in my case with the borked 780. Basically, all you should need to is lower the "power limit" slider from 100% to something like 80%, maybe lower.

realnc wrote:

Undercloking with MSI Afterburner might help. It did in my case with the borked 780. Basically, all you should need to is lower the "power limit" slider from 100% to something like 80%, maybe lower.

Along those same lines, using a bit more voltage may fix it too since it doesn't seem heat related.

It can't be the core I don't think, but memory I could see being the culprit here. But I've never heard of memory errors shutting down systems like that, unless BSODs are preceding said shut downs.
I second the idea to try underclocking and/or overvolting/undervolting using afterburner. I would test sequentially:

1) Lower Core Clock significantly, if it shuts down again return to default and move to the next
2) Lower Memory Clock significantly, if it shuts down again ...
3) Lower Voltage somewhat if possible, if ...
4) Increase voltage by some margin, I don't know the maximum safe for 980 Ti, but it should be easy to find and then move down a little bit and see how it reacts

Overall, pretty weird case, I would also try a clean driver installation with DDU as well. Wondering if Nvidia broke something with somewhat recent driver releases perhaps? Perhaps something to keep in mind, try running drivers from last year or so just to be sure.

What PSU are you using in the computer? Usually when you have a complete computer shutdown like that, it is due to a lack of required current on the main power rails. Could be the games you were playing simply weren't requiring all the GPU had to give, but then other games kick in something that takes the PSU down with it.

Hm, I would say that a GPU stress test will demand more power than games so I don't see problem there.

Maybe a problem in some area of video memory that is not always used. Mem. chips tend to fail during the aging of the card (due to "high quality" ROHS soldering and thermal stress but it usually shows visible artefacts in image. To be sure you can test the VRAM by various tools (at least VMT + e.g. MemtestCL). See links on my page: http://rayer.g6.cz/hardware/gtx670.htm#SGRAM_REBALL.

Hi Guys,

Thanks for the replies! Some good ideas there. The card did the same thing in my friends computer. I have found that around 50% of my games work with no problems, and are usually older titles but not exclusively. Some newer titles perform perfectly without ever shutting down.

I also don't think it's heat related because we have the heat under control using the fan profiles.

I will try the following : Underclocking to the maximum level the software allows, voltages (up and down) and I'll run those VRAM tests that RayeR suggested. If all of this fails, I will try to flash with another bios. Would be a shame to have to bin such a great card! When it runs, it runs like the wind.

I am using a Corsair Builders 650W PSU, almost brand new, no problems with PCI-EX power delivery. It's definitely the card itself. I am also leaning towards a faulty piece of VRAM, given the strange nature of when the problem actually kicks in.

Was playing Civilization 5 at 2K with all details on and 8x MSAA for days, yet starting a game of Civ 6 at 1080p medium settings (no AA) shuts down the PC in minutes.

Gonna try these solutions and get back to you all. Thanks again!

I have only 430W Seasonic PSU +5years old for GTX 670, powered by 2 extra cables and no problem...
I changed link to my webpage with links to test utils.

Cheers RayeR I was a little confused!

All the best for 2019.

I'll keep y'all posted! See if we can't solve this issue.

Maybe some interesting tools.
https://www.raymond.cc/blog/having-problems-w … est-its-memory/

Turned out to be a VRAM issue ... it seems any game that doesn't use more than the first 2GB VRAM usually worked fine without crashing / resetting the PC. As soon as that 3rd (4th?) GB was accessed the PC would hard reset. I used Superposition benchmark and the test would run all day in the lower settings / resolutions, Superposition tells you exactly how much VRAM will be used as you tweak the settings. Whenever the VRAM usage approaches 4GB *BOOM* off it goes.

If only there was a way to turn it into a 3GB card, or even 2GB.

Eventually it fried the MB (while the card itself is still strangely intact). Nothing can be done with it I guess? Might make a handsome ornament.

Fried the mb? How?

Ozzuneoj wrote:

Fried the mb? How?

Not entirely sure!? As I said the card would randomly and sporadically force a hardware reset of the PC ... it seems to have happened whenever that 4th GB VRAM was about to used. I would have to power off the rig and wait a few seconds, then turn it back on and everything would be fine until I tested another game which would use that bad VRAM. Anyhow I was play Civilization 6 to see if I could get an hour or so gaming before it shorted out but after just a few minutes of play, the computer turned off again and this time I couldn't get it to post. Replaced the card with a working spare HD6670 and nothing. Tried the PSU but eventually realized it was the board that had blown. Replaced the board with a spare Gigabyte socket 1155 I had and decided to sack the 980Ti to the parts draw indefinitely. I had put the 980Ti in just to see if it had died in the "accident" but it lived on, just in it's crippled state.