RTX 5090 black screen/freeze crashes
retained text
**RTX 5090 Black Screen Crashes – Everything I’ve Tried So Far** I’ve been experiencing random black screen crashes with my RTX 5090, sometimes accompanied by GPU fans ramping up to 100%, frozen audio and forced shutdowns. **PC Specs:** GPU: MSI RTX 5090 Ventus 3X CPU: Intel i9-14900K Motherboard: ASUS ROG Maximus Z790 Dark Hero RAM: 64GB DDR5 6400 MT/s (XMP I) PSU: Seasonic PRIME PX 1600W OS: Windows 11 **Errors:** WHEA-Logger 17, nvlddmkm 14/153, LiveKernelEvent 117/141. **❌ Things I’ve tried that didn’t fix the issue** Clean NVIDIA driver installation using DDU. Tested multiple NVIDIA driver versions (older and newer). Removed RivaTuner Statistics Server. Updated motherboard BIOS. Updated motherboard/chipset drivers. Removed CPU undervolt and restored stock settings. Disabled various PCIe ASPM and power-saving options. Changed Windows PCIe Link State Power Management. Increased Windows TDR delay to 10 seconds. Ran Windows SFC system integrity checks. Investigated HDMI/DisplayPort and dual-monitor configuration. Tested PCIe Gen 4 instead of Gen 5: mixed results, with temporary stability but no definitive fix. **✅ Things that seem to help / recent positive results** **Physically removed and reseated the RTX 5090** in its PCIe slot. **Removed and properly reseated both RAM sticks**, ensuring the locking clips engaged correctly. **Removed leftover protective plastic film** from the GPU’s rear metal bracket (not confirmed as the cause). **OCCT GPU stress tests:** Multiple successful one-hour runs with zero errors. **OCCT VRAM tests:** No errors detected. **CPU and RAM stress tests:** Passed without reported errors. **MemTest64:** One hour at 6400 MT/s, zero errors. **Latest GPU stress test after reseating components:** Passed for one hour with zero errors. **Extended gaming sessions with OBS running:** Several hours without crashes on some days. **Current situation** The most interesting detail is that **the crashes often happen during low-load situations or when returning to the PC after being idle**, rather than during heavy GPU stress tests. WHEA-Logger Event ID 17 sometimes appears right before NVIDIA driver errors, suggesting a possible PCIe communication issue. After the latest physical inspection, GPU/RAM reseating and additional stress testing, things look promising. **However, it’s still too early to confirm that the issue is permanently fixed.** Has anyone experienced something similar with an RTX 5090? Did reseating the GPU, changing PCIe settings or another solution permanently fix it?
document fetched 2026-10-28 00:32 · canonical https://reddit.com/r/StableDiffusion/comments/1xgornq/rtx_5090_black_screenfreeze_crashes