Cinebench - PC shuts down immediately when I start it
Cinebench - PC shuts down immediately when I start it
Hello everyone, I just upgraded my CPU and GPU and everything is working well so far. I haven’t done any overclocking yet, so I wanted to check Cinebench to compare my performance with my old system. However, whenever I start the tests—whether multi-core or single core—I immediately notice my PC shutting down. It’s important to note that I’m not experiencing any problems with gaming or other tasks; everything runs smoothly.
Do you know why this might be happening? How can we fix or troubleshoot this issue? I’ve shared my specifications and the steps I’ve taken so far.
Specs
Mobo: MSI B450M Mortar
CPU: Ryzen 9 5900x
GPU: MSI RTX 3090 Ventus 3X 24GB GDDR6X
RAM: Kingston Fury Renegade DDR4 3600MHz CL16 32GB Kit (XMP 1)
PSU: Corsair RM 850 W 80+ Gold Fully Modular ATX
Storage: WD SN750 1TB NVMe SSD with Heatsink
What I’ve tried
I’ve tested various Cinebench versions (R23.2, R23, R20).
Removed software like MSI Afterburner, RTSS, Razer Synapse, Ryzen Master, HWInfo, etc., thinking they might cause conflicts.
I ran the PC in Safe Mode and observed random errors during tests; the app also fails to load properly.
Downloaded Cinebench from the Microsoft Store instead of Maxon.
Adjusted Windows power settings from Balanced to Ryzen High Power.
Looking forward to your advice—thanks a lot!
Hey, everything seems fine with 3dmark and no other problems. I actually have a power meter, and the maximum the PC draws under heavy load is about 500-550W, which looks good. The event viewer showed the PC shutting down unexpectedly, but that doesn’t give any more details. Maybe I’m checking the wrong place?
Hey there,
As suggested, I'm considering power. Although the RMx is a solid PSU and often recommended for quality gold units, that doesn't guarantee your setup is functioning well. The RTX series GPUs experience very quick transient power spikes. When this happens, the PSU switches from low to high in just a microsecond, triggering over-protection and causing a system restart. This might be the problem.
Do you still have the previous GPU? Would it help to replace it with something other than the 3090 and test again? Or you could try using a different PSU altogether.
Hey, thanks for the reply!
I do have the old GPU, so definitely could test it out with that, however, the power meter I'm using shows that the power draw is usually at around 500-550W at high load, and no issues running any other benchmarking software like 3dmark. I've also had no issues gaming for extended periods of time - so just wondering if you still think that the PSU/power is likely a culprit here? Wouldn't I also expect to see issues with 3dmark and/or gaming for extended periods of time if power was indeed the issue?
One thing I did realise was that a couple of times Cinebench (the app itself) didn't actually open up at all. Similarly, I noticed that the CPU score was quite low compared to the average for 5900x on 3dmark, so I'm wondering if I'm not utilising it fully for some reason. I haven't made any changes to the out of box settings so far, but I did undervolt my GPU a little.
Because it's Cinebench, your GPU shouldn't exceed its 2D clocks (showing the desktop). Have you verified the RAM placement—should they be in A2 and B2 slots? No issues detected? You can run HCI Memtest or Testmem5, Linpack Extreme, or Y-cruncher (2.5b is a longer test but still brief). Faulty or unstable RAM may also trigger random reboots. Are chipset drivers installed? Cinebench should fit well in your L3 cache, making it a less demanding system test for newer CPUs.
Yes, your typical wattage might be around that level, but the sudden spikes can easily raise it to over 750w. Check the other GPU using Cinebench. If the issue isn't there, it's likely the 3090. Start with that.
It's good to note that your bios should be updated as well; running an MSI MMO and RTX caused problems, but they've been fixed with a bios update. This might be the reason.
But he has an 850W PSU. A quality one at that. It doesn't seem to exceed 900W much. I'm not sure how the OCP is configured on this model, but it should manage spikes well.
I'm not entirely confident about the problem. The next step would be checking the CPU, GPU, and RAM one by one. Test each individually to see if anything fails. Also, verify that the BIOS settings are correct—Vcore, clock speeds, voltages, etc.—are all properly configured. Look for any discrepancies.