Behavior at extremely low temperatures with air cooling
Behavior at extremely low temperatures with air cooling
I'm facing a puzzling situation: I own a fairly new build with 64 GB RAM, 3200 MHz RAM speed, an Aorus Pro Wi-Fi ITX X570 board, and a 3950x with NH-U12A. I haven't overclocked anything, and PBO is set to auto. The room temperature in my office is between 19-20°C. I haven’t removed the CPU heatsink or used any liquid cooling solutions—just the standard thermal paste from the cooler.
I’ve been testing with Prime-95 and performed small FFTs on the CPU. The Task Manager shows maximum usage across all threads, with frequencies ranging from just below base to the high 3000s. This seems normal, but in HWInfo (and Ryzen Master) my CPU, which usually idles around 40°C, is running at 58-62°C under full load. I noticed occasional spikes up to about 70°C, but those still feel unusually low.
My first suspicion was that the temperature sensor might be faulty, but it doesn’t seem to be giving me any misleading readings. I’ve built many PCs that display heat like a space heater and this one behaves similarly.
Next, I considered running stress tests to see if something is wrong with the CPU itself. A benchmark could push a failing chip to its limits even if performance isn’t matching expectations. However, when I ran Geekbench 5 and Cinebench R20, the scores looked normal—temps stayed under 60°C, and maxes were around 63-64°C. Still, the numbers don’t match what I see online.
Someone with experience in similar setups has any insights?
Hello there. Avoid Hwmonitor for Ryzen systems. Ignore that suggestion. Opt for HWiNFO instead. If @rmartin558 is operating near 3.8GHz, the processor likely isn’t drawing much power and temperatures are reasonable—particularly with a 20°C ambient. Never rely on HWmonitor with Ryzen; stick to HWiNFO.
Noctua thermal paste performs well, matching the quality of its standalone products. The appearance and packaging are standard, but the performance feels robust. It resembles a high-quality silicon chip running smoothly under load, with no noticeable differences.
Thanks for your feedback. It seems both tools are showing similar results, which isn’t surprising at the moment. I used HWiNFO before mainly because of my familiarity, and it felt more reliable to me. Your concern about temperatures is valid—higher temps can affect performance. In a scenario with all-core load, the system might naturally adjust to maintain stability. It’s understandable to feel a bit concerned about overclocking, especially if you’re not planning it. Let me know if you’d like tips on monitoring or managing temperatures better.
Adjust the PBO options or try the EDC bug for extra performance, though expect increased temperatures. The ambient air inside the case contributes significantly, which is beneficial. A warmer room by about 7°C would clearly affect the CPU readings.
it uses a large database of AMD’s preferred settings. the processor evaluates how it performs with various numbers of cores, temperatures, and power usage, then selects suitable values from the chart to maintain efficiency. you can experiment with fewer threads than total available and observe if performance improves. each core has its own frequency range depending on factors like temperature, power draw, and duration at that frequency.