Power off abruptly and reinitialize BIOS during intense gaming or rendering tasks.
Power off abruptly and reinitialize BIOS during intense gaming or rendering tasks.
Hi! I've been experiencing a weird and very annoying trouble with my PC. My machine shuts down whenever I load it, doesn't matter if it's a gaming load or a work load (Vray - CPU based rendering). It usually happens after 3 to 4 hours of use. It's often a hard shutdown with the characteristic click coming from the PSU, the pwr led stays on but everything else seems powerless. Sometimes (like 3/10 times) the bios goes back to default settings, but I know the mobo is not the problem as it's brand new. I replaced the first one because of the same problems. In terms of troubleshooting, I stress-tested the ram for hours and the ssd's are healthy as well, there's no trouble there. The mobo can be discarded, i guess. This means that the GPU, CPU or PSU might be the problem. I know CPUs don't fail very often if they are not mistreated (overheating, too much voltage, overclocking) which is why I think that is not the issue here. The fact the the shutdown happends regardless of GPU or CPU load and the loud click leads me to believe that there's a PSU problem. Note: CPU and GPU temps are constantly checked and everything is normal. The only thing that seems a little bit odd is that now CPU clocks under load are a bit too low sometimes. Usually when I was rendering clocks wouldn't go below 3.9, now im seeing 3.7 and even 3.6. I'm relatively new to PC's, soo... I could be wrong. Any suggestions will be greatly appreciated. Hardware: 3900x Gigabyte x570i Aourus Pro WiFi 32GB (2x16) 3200 C16 Patriot Viper Corsair sf600 RTX 2080 Black - EVGA
Compact power supplies are known for struggling when under heavy load. A major factor in why I rarely suggest using a 3080 or higher in a SF build is that these units simply can't cope with them, even the 850W SF models may have trouble. While small form factor offers some advantages, you need to be very aware of its drawbacks. You might manage an 850W SF PSU with a 2080, but beyond that the limits will set in—power consumption will rise significantly, and the PSU will likely need to adapt to handle more sudden spikes from the GPU, which can trigger hard shutdowns. The next generation of PSUs should be more capable, though this won't be available until after the upcoming GPU release, probably within a few months.
The latest BIOS version is available, but the power supply unit is relatively low for this model.
At least several of these scenarios (you're not the first) likely stem from RAM that doesn't meet the board's requirements. Initially, I should verify if your RAM model appears in the manufacturer's QVL for your specific board and processor—yes, the processor can influence this. Some argue QVL is irrelevant since RAM is RAM, but my observations differ. Occasionally you might find RAM not listed that prevents optimal performance, yet stability, reliability, and compatibility remain uncertain. Not everyone is aware of this, but Ryzens (possibly even AMD's own chipsets) are quite particular about RAM. After four months of troubleshooting random no-video POSTs after a hard shutdown during gaming, I had to return the board with an RMA. During that period, I encountered similar problems with the same DIMMs on another board, reinforcing that the GPU wasn't the problem. One DIMM failed to boot at all. The board was marked as "could not duplicate issue," and I learned the DIMMs weren't in QVL. I swapped them with RAM from each board's corresponding QVL, and both systems worked smoothly. I haven’t faced comparable issues with the same components on an older machine, but I’m confident those stability concerns will be resolved similarly.
Have you checked the power phase temperatures on the motherboard? This board features a solid 6+2 VRM setup, but even in a smaller ITX case, if CPU turbo slows down after long sessions, the VRMs might be getting warm. Could you tell me what case or chassis you're using? Also, how is the airflow looking—especially around the CPU socket? And what type of cooling solution are you employing?
The previous version was F36a. It seems this concern has existed for quite some time—I've likely updated through 3 to 4 BIOS releases before noticing the issue. I placed a new PSU just to verify it works, and thank you for your patience.
I completely understand your point. It makes sense if a brand compiles a large list of compatible RAM, likely for a specific reason. I remember at the time that compatibility with Ryzen was important, but I don’t recall checking it during setup. After fixing the PSU, I thought testing RAM might be next since I could easily find a cheap 4GB option. Thanks for the feedback!
Absolutely, this VRM is quite effective. I reviewed it some time back and everything was fine, but I'll recheck those temperatures just to be sure. The setup: Louqe Ghost S1 Cooler with Noctua NH-L12 (GS1 Ed.) It never exceeds 86°C during rendering. This cooler works great for VRMs since it directs air straight to them. Thanks for the tip!
After thorough testing... The problem was traced to the power supply unit. I examined the GPU power usage graph in gpu-z and noticed shutdowns linked to a particular task that caused extreme load, pushing TDP to 99%. I recreated the same situation using another PSU multiple times, and the issue disappeared. I used a 750W EVGA and it performed perfectly. Thanks to all the helpers! See you soon!