System runs slowly on startup and frequently freezes.
System runs slowly on startup and frequently freezes.
Here’s a revised version of your text:
The issue involves several crashes under heavy GPU usage, especially during AI tasks or gaming. After troubleshooting, I removed all mods and resets in games, then reinstalled the latest NVIDIA drivers using a DDU. I tried it for 15 minutes without any problems. A 4-hour memory test from a USB drive passed smoothly, and I ran OCCT to stress-test the GPU for an hour without issues. I also tested VRAM, 3D Standard, and 3D adaptive features using OCCT, which reported no errors. The pass came after a thorough check.
I ran OCCT to evaluate GPU performance, including VRAM usage and CPU stress tests. It showed stable results for 1 hour. I used prime95 to stress-test the CPU, which is currently running at 78°C with no issues.
In my setup, I have Windows 10 Pro, the newest NVIDIA drivers for the 3080 graphics card, a 10GB SSD, 32GB RAM, and an X570 AM4 motherboard. The BIOS is up to date. I’ve reseated the drives and GPU multiple times, updated the BIOS, and replaced the hard drives. Initially, everything worked well for a while.
Recently, I experienced intermittent crashes with the same error message (WHCEA uncorrectable). I suspect it might be related to PCIe lanes, as previous reseating helped. I’ve also been reseating the GPU and motherboard, but it’s still unstable.
My thoughts are that either the motherboard is faulty (I can replace it), or the GPU is defective (I plan to upgrade because 10GB isn’t enough anymore). I’ve tested the GPU with VramTest and it passed 5 minutes, but longer tests are needed. I’ve also reseated the drives several times and updated everything.
Currently, after another crash, the system boots very slowly—about 40 seconds now compared to a few minutes before. Other than that, everything seems normal, and I can still play games.
Your GPU is already set to a lower voltage, which helps prevent overheating. The BIOS adjustment for PBO should not affect boot time directly, but ensure the system is properly configured for optimal performance.
As a tower setup, what weight are you expecting between the CPU cooler and GPU? I've noticed cases where microcracks form in the board, initially causing problems as heat builds up and causes incomplete connections. Over time it worsens, similar to what you're facing. I'm proposing an unusual approach—place the case flat so MOBO is horizontal and the cards upright—to test for differences. This might be a bit unconventional, but I think people often overlook the stress the weight applies to components during troubleshooting.
CPU cooler is completely lightweight, it's an AIO design. The 3080 isn't the biggest problem—it's just a small 300mm card, likely from a Dell OEM model. If the motherboard is faulty, I'll go ahead and replace it. It's better to spend 300 CAD on a new CPU and RAM rather than wasting good components. I might consider that option.
I eliminated PBO and the circuit remains consistently undervolted.
Turning it off now will make things slower but okay. If I can find the issue, I can invest in fixing it. RAM is affordable, so I wouldn’t mind replacing it.