No, these temperatures and PBO limits are not safe.
No, these temperatures and PBO limits are not safe.
CPU cooler recommendation is the Corsair - ICUE H100i Elite Capellix 240mm with thermal Grizzley. I lowered all cores by 30 degrees and set PBO limits to unlimited. After a 10-minute Cinebench run, the temperatures were as follows. What worries me is that after reaching 84°C, the drop seems slow—about 3°C per second instead of what I expected. I applied a thin paste and ensured proper airflow. I also have an MSI - MAG CORELIQUID 360R V2 radiator, but the temps felt similar during testing. Who can advise which cooler performs better technically? And if the paste application isn’t ideal, would heat dissipation to the copper plate be slower in terms of cooling speed?
You've experienced no power issues or unusual behavior. To verify stability, you could run cinebench r23 repeatedly and check if temperatures remain consistent. If you're considering upgrading, the MSI MAG CORELIQUID 360R V2 radiator with a built-in water pump might be a better fit compared to the Corsair H100i, which uses a separate block. Your current MSI cooler in an MSI MPG gunr choker case seems to perform similarly, but switching cases could affect performance.
Based on my observations over the last year and a half, Cinebench usually didn’t trigger issues. Problems tend to appear only during high load or specific games—like freezes, restarts, or BSODs. The cooler might drop a bit with the 360 model, but under full load it performs well and stays stable. This topic keeps coming up in discussions about Ryzen 5000, and I found a helpful thread here: https://old.reddit.com/r/Amd/comments/js...3_thermal/
It's a common approach to evaluate thermal behavior once adjustments are made using PBO and Curve Optimizer.
It's unusual for low load conditions to reveal issues with the BSOD. I attempted to overclock using undervolting and followed a YouTube tutorial that set PBO limits to unlimited at -30. However, I found it more stable at -20 in my experience. There was a crash during a YouTube video even though benchmarking tools showed stability. It might be better to retain the H100i as it functions well as an exhaust, helping keep GPU temperatures low when the 360mm was used as an intake.
It takes a lot of time to pinpoint each core limit by testing how far you can go in the CO, so you probably bypassed that step and jumped straight to -30. The tool I discovered for checking each core separately and spotting issues is corecycler. You can read more about it on its GitHub page.