My ZOTAC GeForce GTX 1080 Ti AMP 11GB recently began crashing in several games, including Valorant, Counter-Strike, Black Myth: Wukong, and Hitman. Depending on the game, it either closes to the desktop, freezes, becomes unresponsive, or occasionally locks up the entire PC. Windows logs repeatedly show nvlddmkm Event IDs 13, 14, and 153, including "Illegal Instruction Encoding," "Physical Multiple Warp Errors," GPU errors, and previous LiveKernelEvent 141 reports with an Error_DMA_PageFault and graphics engine reset.
The card worked reliably for about a year in this system. The hardware is an Intel i5-10400F, Gigabyte H410M S2 V2 motherboard, 16GB DDR4, MSI MAG A650BN 650W PSU, Windows 11, and the GTX 1080 Ti. As an A/B test, I replaced the 1080 Ti with a GeForce GT 710 in the same PC. The games ran without these crashes, and reinstalling the 1080 Ti caused the problem to return.
I have already replaced the PSU and system drive, reinstalled Windows, used DDU with multiple NVIDIA drivers, updated the motherboard BIOS, installed chipset drivers, passed MemTest86 and storage checks, tested PCIe Gen 2 and Gen 3, confirmed the card runs at PCIe 3.0 x16, disabled ASPM and virtual display software, and checked for missing devices. Lowering the GTX 1080 Ti to an 80% power limit, -150 MHz core, and -500 MHz memory did not solve the crashes. Short OCCT VRAM and general tests, Unigine Superposition, and Premiere Pro GPU rendering can complete successfully.
Do the warp and illegal-instruction errors suggest failing GPU silicon, or could faulty GDDR5X VRAM cause them? How reliable are short VRAM stress tests, and is there a useful diagnostic that can distinguish GPU core, VRAM, PCIe, motherboard, CPU controller, and driver problems? Given that the GT 710 works in the exact same system, is the 1080 Ti itself the most likely failure point?
2 Answers
This sounds strongly like a failing card, with VRAM being a particularly plausible culprit. Older GTX 10-series cards can develop memory problems that only appear under specific game workloads, so passing a short VRAM test or a benchmark does not completely rule that out. The fact that the problem follows the 1080 Ti while the GT 710 works in the same motherboard, operating system, and power setup is a much stronger piece of evidence than any single stress test.
If reducing the memory clock by 500 MHz does not make the games stable, there may not be a practical software fix. Watch for artifacts, corrupted textures, or freezes where audio and game processing continue briefly, since those symptoms can also point toward memory instability. A long VRAM-focused test may provide more evidence, but a clean result still would not prove the card is healthy.
The progression described is consistent with a 1080 Ti reaching the end of its life. VRAM faults can initially appear only in games and may later become obvious as artifacting or complete failure. GPU driver events such as these are generic recovery reports, so they do not independently identify the exact defective component, but repeated errors across multiple games and the successful replacement-card test make a driver or motherboard problem much less likely.
The most convincing next step would be testing the 1080 Ti in another known-good system, or testing another similarly powerful card in this one. If the 1080 Ti reproduces the crashes elsewhere, treat it as faulty. At that point, repair or replacement is more realistic than continuing to adjust BIOS, PCIe, or driver settings.
Related Questions
Lenovo Thinkpad Stuck In Update Loop Install FilterDriverU2_Reload