My Lenovo Legion Y540-15IRH has an RTX 2060 6GB, Intel i7-9750H, and 16GB of RAM. After a graphics-driver update, it began freezing during games with colorful square artifacts and repeated symbols across both displays. Sometimes the whole system crashes with a BSOD, while other times only the game crashes and the screen flickers. The problem initially appeared occasionally but has become reproducible within about 15 minutes of loading new 3D textures. Most 2D games remain stable, and the GPU can sustain heavy workloads when no new assets are being loaded. Warframe is an occasional exception. Event Viewer repeatedly reports NVIDIA driver errors involving nvlddmkm.sys, especially Event ID 153, along with IDs 14 and 13. Most minidumps mention a kernel-mode driver accessing pageable memory at an elevated IRQL; another references a fatal video scheduler violation. I have tried several driver versions, DDU removal and clean reinstallations, Windows and Lenovo updates, resetting BIOS settings, disabling audio devices, disconnecting the second display, and other software cleanup. None permanently helped. Resetting the BIOS temporarily restored stability for several hours, but the crashes returned after playing another demanding game. The laptop has not obviously overheated, and the GPU is soldered to the motherboard, so replacing it would be difficult and expensive. Could this still be a failing GPU, motherboard component, memory, or cooling system? Are there worthwhile diagnostic steps before trying an underclock or replacing hardware?
3 Answers
A system that behaves normally for a few hours after a BIOS reset and then develops stuttering again could be experiencing throttling, unstable power delivery, cooling problems, or a failing component on the motherboard. Heavy stress testing is not conclusive if the fault is triggered specifically when new textures are transferred into video memory. Check GPU and CPU temperatures, clocks, power limits, and whether the crashes correlate with VRAM usage. Also test system RAM and run the laptop with conservative performance settings. The laptop version of the RTX 2060 is a mobile variant, even though Windows and many utilities commonly display it simply as RTX 2060; it is not the same power configuration as the desktop card.
The minidumps are useful, but tools such as BlueScreenView can help compare all crashes and show whether the same modules appear repeatedly. Seeing nvlddmkm.sys alongside ntoskrnl.exe usually points to the NVIDIA graphics stack, though it does not by itself prove the driver is defective. The separate dxgkrnl.sys, dxgmms2.sys, and watchdog.sys crash may indicate a broader graphics or video-memory problem. Since the laptop is several years old, hardware degradation should remain on the list even if software has also been ruled out.
Because the artifacts appear before the BSOD and the failures happen while loading textures, VRAM or the GPU's memory interface is a serious possibility. Driver cleanup and rollbacks are still reasonable, but repeatedly copying nvlddmkm.sys or changing unrelated drivers will not repair a defective graphics component. Before concluding that, test with a clean Windows installation or a known-stable driver, disable overlays, and use a mild GPU memory/core underclock as a diagnostic. If lowering clocks or power makes the crashes disappear, that strongly suggests marginal hardware or power delivery rather than a normal driver problem.
The BIOS reset temporarily making the system stable is what makes me unsure about hardware failure. If the GPU or its surrounding circuitry is failing, would that normally require replacing the whole motherboard, or could a repair shop address a smaller component?