1080 ti crashing under load but not during benchmarks
A few months ago I upgraded to a GTX 1080 ti (considered used but acceptable by the seller) for work and personal purposes.
I have since had a couple issues with it, not least recently, which I’ll detail here:
Under load during games (my usual testing game is the Long Dark) the graphics card crashes. I still get audio feedback from the game in the background, and can move and access menus. When this happens, MSIafterburner and HWinfo show a sharp decrease in activity (about 1%).
I also get artifacting sometimes, typically before a crash.
I was also getting BSOD, but this has been fixed by BIOs updates in all relevant cases
My specs are as follows:
Processor: AMD Ryzen 5 3600 6-Core Processor, 3600 Mhz, 6 Core(s), 12 Logical Processor(s)
Motherboard: X470 GAMING PLUS MAX (MS-7B79)
GPU: Nvidia GTX 1080 Ti
RAM: Ballistix 16gb 3600 mhz DDR4 (x2)
PSU: XPG CORE Reactor 750Watt 80 Plus Gold Certified Fully Modular Power Supply (COREREACTOR750G-BKCUS)
I have attempted:
Reinstalling windows (which appeared to work for about a week – though this was to fix the BSOD issues (Which it didn’t even do))
Uninstalling and reinstalling the most recent (3080 game ready) drivers
Fully uninstalling ALL drivers and reinstalling relevant ones
Uninstalling PhysX (which I then reinstalled after the crashing continued)
Fully uninstalling ALL drivers and rolling back to an earlier set
Run benchmarks on GPU using Furmark and Heaven Benchmark 4.0 (Heaven did crash once, but only because I plugged in my wireless headset during the benchmark – on subsequent attempts it did occasionally drop frames).
Monitored temps during tests (which remained within suitable levels)
Literally upgraded my PSU from a bronze rates 750 W to a gold rated
Ran CPU and memory benchmarks using aida64EXTREME (no errors I could see)
Updated BIOS on my old motherboard
Literally replaced the entire motherboard
Updated BIOS on a new motherboard (to attempt to see if it was a fault with the single PCIe I had on my old motherboard)
Checked BIOS for overclock profiles (none existed)
Ran memtest x86 on both my RAM sticks (no fails or errors)
Run BenchmarkTest on entire setup (regular functioning across the board)
Underclocking using MSI afterburner
Increasing core voltage to my graphics card
Reseating the graphics card
Rolled back windows updates
Monitored temperatures pre-crash (regular, 50 c temp)