I'm experiencing issues with my RTX 5090 where it crashes under heavy GPU load, and I'm trying to figure out if this is a hardware defect or a software problem. Here are my system specs: Intel Core Ultra 9 285K, RTX 5090 32GB, MSI PRO Z890-S WiFi, 64GB DDR5 5600 MHz, 1050W Gold PSU, and I'm on Windows 11. The crashes only happen during demanding tasks, like ray tracing games and AI workloads. I'm getting full shutdowns and screen artifacts right before these crashes occur. What's puzzling is that under light usage, everything runs perfectly fine. Here's what I've already done: a clean Windows 11 install, tried multiple NVIDIA drivers, reset BIOS settings to default, and checked all connections. I've also monitored GPU temperature and power usage during these crashes, and everything seems normal. I'm starting to suspect a hardware issue either with the GPU or the VRAM itself. Does this sound like a likely hardware failure? Are there any known VRAM stability issues with the RTX 5090, or specific stress tests I should perform before sending it for repairs? Thanks for your help!
4 Answers
Have you tried updating your BIOS? Also, reseating your RAM and GPU might help. I had stability problems before, and a simple BIOS update fixed everything for my setup. Sometimes, it’s not just about the components but also how they’re configured.
I created a public stress test repository that mimics the workload patterns you’re facing. It reliably crashes my RTX 5090 after about 2 minutes under sustained load. You can find it [here](https://github.com/ricoxor/stress-test-ollama-rtx5090). If you try it out, please let me know if you experience crashes and which NVIDIA driver version you're using!
You should consider testing by removing half of your RAM. Sometimes it could just be bad RAM, or it might be pushed too hard with the settings. Just make sure that the RAM is properly seated in the correct slots.
It sounds like your GPU might be facing some stability issues, but don’t rule out the RAM just yet. Since you’re running at 5600 MT/s, try downclocking it a bit to see if that helps. I had similar crashing with my setup until I reduced my RAM speed. It’s worth checking your memory stability along with the GPU.
Thanks for the advice! I'll definitely try downclocking the RAM and run some stress tests.