New 1080ti = DCS Crashing, help needed

I bought a new Gigabyte Aorus 1080ti Xtreme Edition on Tuesday. Unfortunately DCS, BoS, Assetto Corsa crash almost immediately - when in intensive situations, like approaching Normandy coast, or flying over Las Vegas. I have tried different drivers, no change. I tried changing the profile in the Gigabyte software from "game mode" to "silent mode", which underclocks the card, and it was more stable, and didn't crash. When it crashes I get lots of squares on the screen, like a broken image. The computer keeps working, and sound sometimes continues, but the game locks up and quits.

I checked my temperature in GPU-Z, and it is low (46-60) because the game has only been running a short time. I took it back to the retailer and they stress tested with FurMark and Unigine Valley, both of which actually didn't crash. Today I replaced my PSU with a 750watt Corsair Gold X, and still have the issue, so it's not PSU related. All I know is that I didn't get any issues with my 1070 earlier in the week.

I have attached my log files, can anyone help figure out what it might be? My gut feeling is the 1080ti is faulty in some way, but it's hard to prove to the retailer when their testing doesn't crash the machine.

dcs.log-20180103-113831.zip

If you have not already done so, I would start by doing a clean driver install, using DDU ( Display Driver Uninstaller).

Run DDU in safe mode, deleting the Nvidia drivers.

Grab the latest Nvidia drivers from Nvidia.com, and clean install only the driver itself and PhysX. Do not install GeForce Experience or any of that other stuff.

Reboot the PC, and give DCS another go.

If same problem still persists, I would be inclined to believe it is the card itself.

same with my pals gf980ti, crashes on his, works flawless in mine.

DDU in safe mode & reinstall

Thanks for getting back to me guys - yes, I used DDU on the three iterations of drivers that I have tried, including the latest ones, as well as 384.94 and 388.59. I made sure I was in Safe Mode as well. Given the problem is less apparent when I underclock the card, it's got to be the card right? Considering I replaced the PSU today and ruled that out. Is that grounds for RMA?

Yep, sounds like a weak card to me. I would push for a replacement.

I just got it to crash repeatedly in PassMark, during the 3D rendering tests, multiple tests locked up.

I would start looking at RMA-ing the card. But, before you do so, try uninstalling the Gigabyte Aorus overclock engine/utility and use MSI afterburner instead. When I had my 1080ti Aorus (non-extreme) it was unstable using the Gigabyte utility right out of the box, so I switched to Afterburner and it solved the problem. The issue was that Aorus Engine was applying voltage/speeds that card could not handle.

That being said, I still returned my card (thumbs up for Amazon) and picked up an EVGA reference card for watercooling. Haven't looked back.

I just got it to crash repeatedly in PassMark, during the 3D rendering tests, multiple tests locked up.

-->RMA !

damn high end stuff :(

edt:*

doing exact same thing with my old 980gtx, fans failed, since my son's rig only has AIO H110 he needs the fans for now. Since it is still under warranty I hope that my RMA goes smooth. So far its been a pain tbh.

Is your cpu overclocked in any way. I've had this kind of thing happen when I increase cpu Ghz. With or without increasing voltage. Its also happened when I bought ram that has the clock numbers // 9-0-0-1 (whatever) mismatched slightly.

No, my CPU is stock 4ghz with 4.4 boost, and aftermarket cooler. The whole system is otherwise stable and cool. In PassMark, only the 3D tests fail, and they fail constantly, in all tests. Never had a single crash with my factory overclocked 1070.

For what its worth, the latest 2.2 update freezes permanently for me every 20-30 minutes, but is buttersmooth on Normandy. So I wouldn't use it as a benchmark. Previous version was stable, but had stutters.

Your card is defective, if you can reproduce the crashes with Passmark. there is almost zero doubt about it.

I would not try to fix it myself in any way but go for RMA.

Try it in a different PCIe slot on the mobo if possible.

So the retailer has requested I leave my PC with them so they can test it. I asked them to test in the same conditions, but my fear is they will run a couple of benchmarks on a 1080p monitor, which won’t stress the card. I can prove the card is the issue, it’s frustrating that, that alone isn’t enough.

Cant you possibly say you need your PC (and your old card) for work?

Well, this still leaves the chance they dont stress it enough to make it crash and reproduce the issue... :/

So the retailer has requested I leave my PC with them so they can test it. I asked them to test in the same conditions, but my fear is they will run a couple of benchmarks on a 1080p monitor, which won’t stress the card. I can prove the card is the issue, it’s frustrating that, that alone isn’t enough.

I would never leave it with someone.

Sounds like you do not need to purchase from them anymore to me.

Can't you contact Gigabyte and get an RMA from them for a replacement?

that sux :(

you could agree to stay with your pc while you all do tests at their shop. you may carry your screen along.

man, doing same thing with my old 980, 10 days and no "send in" number yet. if it wont come by monday i will call amazon and ask them for help. wish i could send to asus directly but they dont do that for gpu, only srv and ws boards :( baaad

So the retailer has requested I leave my PC with them so they can test it. I asked them to test in the same conditions, but my fear is they will run a couple of benchmarks on a 1080p monitor, which won’t stress the card. I can prove the card is the issue, it’s frustrating that, that alone isn’t enough.

tHE version of that 1080ti is well known to be bugged, just search on internet about it.

That's why I bought the asus strix 1080ti instead of the gigabyte.

You could try removing the cooling fans and heat-sink to visually inspect the card. It may be something on the PCB board is shorting out or you're getting an intermittent bypass somewhere. If you're patient the safest thing to do is RMA it and you can always ask for them to send you a replacement card before RMAing it so that you can do a direct swap. Tell them to put a hold on your credit card for the full amount.. I've done this in the past with many components and most warranty departments don't have any issue with it. Note that putting a hold on your card DOES NOT make a charge to the card. It just allows them to instantly charge your credit card if you don't sent the damaged video card back. Asus does it all the time so I don't see why Gigabyte wouldn't.

I was thinking that to be absolutely sure you should test the card on another PC. The fault could be elsewhere like the PSU no longer able to meet the supply specs with age. They do degrade over time. 1080Ti's require 250W at load, TBH I wouldn't feel safe with just an old 650W unit.

it can pull a lot more if you chase it. The 250w rating btw is for Nvidia FE default clocks, not what Asus EVGA MSI GiGABYTE sell you with factory overclocked cards, they easily run towards 265-285w. Add your own oc and you have peaks of 300w+, if you want it that hard.

I have seen 2 PSU fail with my buddies new 1080Ti. It also had problems with the 980Ti, also a 250w card, also 1 x 65ow PSU no-name and then a low-tier 1000w, lasted less than 24h with 2 cards. Now he runs a better PSU and has no more errors. Took him 2 PSUs to learn that lesson.

I am not saying it IS the culprit, but a 650w PSU and a 1080Ti is on the edge if you oc anything or the PSU is of lesser quality due to manufacturing or age