I was having this exact same problem. I had to completely redesign the airflow to get it to stop.
2 Likes
Those gigabyte cards are OCed almost to the the ceiling of their limits, at factory settings. I would suggest a manual fan curve topping out 100% fan power at 70c. That way you should see a maximum of under 65c under heaviest load.
Automatic fan control on the aorus is not capable of handling heavy load games. Especially when the chip is reaching the limits and becomes unstable in higher temp
2 Likes
I would also check with @Enopho as he has 2 aorus 1080ti systems. 
I have one.
Perfectly stable in automatic fan control, but under a heavy load the temp would reach around 75 and would become unstable. This card really needs an aggressive fan curve.
2 Likes
Made aggressive curve, GPU @67 under load same thing 
frozen HMD screen reboot required before PC grinds to a halt 
edited
WTF just noticed trying furmark I set fan curve to 100% at 70degrees watched furmark and it just kept going up and up… at 80 i stopped it and then realised…my GPU fans are not activating… aaaaaaaaaaaaaargh.
Set fans to manual 100% and furmark doesn’t go above 63 degrees so gonna try a few circuits of project cars with fans full on and see what happens my god this gigabyte software is buggy
4 Likes
That will do it alright.
5 Likes
those cards are heavily overclocked,
having an extra 10c, 15c could possibly turn the perfectly stable card that’s stable at 65c unable at 75c, 80c.
my fan curve is a straight line from 45c at 0% fan to 70c at 100% fan. works well for me.
abit of noise when gaming but better than overwatch crashing mid match lol
3 Likes
So far so good man, Just had a few circuits of project cars and no hangs GPU didn’t go over 63, I’ll just not have to trust the software and put my fans to 100% when gaming, I didn’t notice they weren’t coming on because I’m always in VR with headphones on. Thanks for suggesting I look again at that 
3 Likes
i have noticed mine to cause a crash at 75c.
at factory core settings with some memory OC.
for me its the core clock and temp causing the unstability, my core is right at the limit, even a 10mhz increase casuses crash at load.
hope it works out for you
oh yeah one more thing, use msi afterburner, not the gigabyte software. msi afterburner is the gold standard for nvidia cards.
3 Likes
my 2 1080ti systems are both stable without changing the default settings. admittedly i play on a 4k not a 5k+ or 8K
1 Like
One thing. Do You tried a stress test for GPU?
I tried it, several times, and my computer has a 99% os stability. But when I start Pitool and pi_server… after a while it crashes.
1 Like
3dmark? I found that was not enough of stress test to be fully stable.
Overwatch is actually quite good for testing instability GPU. Crash happens quick. I play and test same time.
But my stress testing were done for long time. System has been rock solid for ages, no crashes unless it’s game bug.
That said the pitool 91, 95, beta, I’m not getting any crashes at all. The beta has gotta considerably more reliable.
1 Like
Perjaps Unigene’s Susposition(sp?) Might be good to stress test.
For the ultimate stress test, run Super Pi and Furmark at the same time
2 Likes
When I was stress testing my MSI RTX 2080’s overclocking ability, the first (and only) app to have problems was 3D Mark.
Fwiw, my card is a great overclocker (stable with gpu at 1950 Hz (+29%) and mem at 8000 Hz (+14%)). I don’t usually run at those settings; I normally back them down a bit. With the video card’s 3 fans, I’ve never seen it hit 100% fan speed, nor has it ever exceeded 65° C.
1 Like
I think it depends on what game.
Sometimes a game would run it 60c under auto fan, and then the next game will really thrash it at 80c with auto fan.
So I went with an custom curve to completely get rid of this problem.
I test my system ram with a software called ramtest, quite handy, my GPU with 3dmark first 99% then some benchmark loops then finally real world gaming.
CPU with some burn test and finally check system event log for WHEA error.
Whole process takes ages but you end with a system you know is stable. Much easier to isolate potential problems down the road
2 Likes
How did you end up checking for the GPU ram OC stability? This one I ran benchmarks until it shows a consistent performance decrease when increasing the memory mhz indicating error correction are correcting errors causing decrease of performance
1 Like
I did a whole range of performance testing. I saw this effect also. I found the sweet-spot of performance to be 4x the gpu speed, plus 100 MHz. I got miniscule performance gains at even higher ram clocks, but I felt it would be best not to overstress the memory.
1 Like
the same happend to me after messing with pitool beta FFR, my Asus Strix 2080 TI is dying even in 2D games right now where temps are low like 60 Cel. and 1400mhz core clock
I’m also thinking that Pimax broke my gfx card.
Tryied different nvidia drivers ( at least 3 including yesterday drivers) , clean installation , and that not helped. Pimax is disconnected ( DP and USB pluged off)
No OC on card. Damn
RMA or any other idea ? It just happend 1 days after .103 firmware installation
1 Like
have you tried DDU?
this software completely remove all of your old gpu drivers so you can do a fresh install.
I always use this to get a fresh install of graphics drivers.
This has many times solved manyof the weird graphics card problem for me in the past and
has always been a highly recommended software.
1 Like