Displays become dark, graphics cards run at full speed
Displays become dark, graphics cards run at full speed
Hello everyone – thanks for taking the time to go through my case and assist in resolving this persistent problem. About a year prior, I replaced my CPU from a Ryzen 3600 to a Ryzen 5950X. Along with the new chip, I carried all previous components (Nvidia 2070 Super; 32GB of 3600 RAM; 1TB NVMe SSD; additional HDDs; case, etc.). I kept the power supply intact, upgrading from a 550W to a 1200W model from be quiet!, and upgraded my motherboard from a B450 Tomahawk (MSI) to a Gigabyte B550X DS3H. This setup was built for professional use, allowing full use of the 5950X’s extra cores.
After the switch, I quickly encountered an unusual issue. Under heavy tasks (not just synthetic tests), I repeatedly tried to recreate this problem with controlled workloads but failed to find a consistent fix. Once, the graphics card would halt video output, spin fans at full speed, and freeze the whole system. The only solution was switching the PSU off for about 30 seconds before turning it back on. This was frustrating, especially since it happened while working on a project and required a restart.
Concerned about the root cause, I upgraded the BIOS and thought it might be a loose connection between my motherboard and GPU. I reset everything and the issue disappeared temporarily. A few months later, it returned—this time while gaming. I performed a full Windows reinstall just to be safe. After DDUing the system, I continued testing. The problem persisted repeatedly.
Eventually, during a game session, it occurred again. This time, I was more cautious and switched the motherboard back to the original B450 Tomahawk. The issue lessened for a couple of months. However, after some time, it reappeared while working. I suspected either hardware failure on the motherboard or a faulty unit. Eventually, I replaced the board with the old B450 model. For a short period, the problem stopped.
Later, while handling work tasks, it happened once more. This time, the B450 Tomahawk displayed an indicator light—unexpectedly, the CPU light was lit, suggesting a hardware fault. I couldn’t pinpoint the exact cause. To help, I logged HWinfo data each time I ran my model. Today, it happened twice in quick succession while I was both running the model and watching YouTube. The system crashed during the most demanding task. After a hard reset via the PSU, the crash stopped. I attached the log for anyone who might recognize the pattern.
Right now, I’m unsure what to do. Possible fixes include tightening CPU mounting or testing the chip itself. It’s puzzling why it works so well most of the time but fails intermittently. I’ve seen similar reports online and hope someone can assist me in finding a resolution. Thank you for reading my update! HWInfo Log.CSV
Gathering the essential details first...
- Chipset drivers from AMD or the motherboard makers? The AMD versions address black screens and BSODs, they’re more recent than those from MBO suppliers.
- Latest BIOS updates applied to the motherboards?
- Is XMP (DOCP) enabled at a specific frequency? Similar results without it?
- Have you made any adjustments like turning off Multiplane Overlay MPO? This resolves some problems.
- What upgrades are on the CPU? Is there sufficient airflow over the VRM heatsinks? Temperatures during load?
- Did you also replace the PSU cables when switching power supplies?
- Were custom cables, extensions, or GPU risers used?
- Are the GPUs running the latest firmware (vBios)?
- Was the CPU inspected for bent pins?
- Has the system crashed during stress tests like Prime95, OCCT, Furmark, 3DMark Stability, Cinebench?
- Are the SSDs updated with the newest firmware versions?
I'll download chipset drivers Yes Yes, 3600, has happened with and without XMP No, never done that Cooler is the NH-U12S. I have never hit CPU temps of over 85 degrees (that was under load). I"m not sure about VRAM, but the fan is oriented properly and there is air going through. Continuing this as a possibility, the model step that causes the crash most of the time causes the CPU to rapidly consumes high amounts of power and back down to very little, over and over again, until the system occasionally crashes. I did replace the cables No custom camples Unsure, I'll try to flash that as a troubleshooting step No bent pins Does not crash on any of those applications, only been Victoria 3, Risk of Rain 2, and most often running a Python model inside ArcPro No, will check and update though
Updated chipset drivers and SSD firmware. No official Nvidia tool found for vBios updates, but I have the latest driver from 2022-02-08. Also, I removed the side panel to verify the VRMs remain cool. Thanks for the support!
The 2070 Super could be a founder model or an AIB card with unique cooling, VRMs, or design. Have you tried tools like IOBit Driver Booster to verify other driver updates?
It's an FE, I didn't attempt it before. I'll try again next time. Appreciate the help!
Here are some firmware updates available for the founders cards:
- [Resizable Bar Firmware Update Tool](https://nvidia.custhelp.com/app/answers/...pdate-tool)
- [GPU Firmware Update Tool for Display IDs](https://nvidia.custhelp.com/app/answers/...-displayid)
It seems we might want to hold off and observe further developments.
I've experienced my initial BSOD today, indicating a memory management problem. This might be linked to the chipset driver update.