I had the same problem with my Ryzen 5 3600x. Turned out that updating my BIOS and especially the chipset drivers vastly improved my temperatures. The config that came with my board (MSI b450 Tomahawk MAX) was pumping voltage too high.
It won't work miracles; the stock cooler isn't great, and if you have restricted airflow in your case then you still want to address that. But I'm blown away by how much difference I got from software in this instance. AMD's implementation of its boost mechanism on Ryzen 2 (and/or its interaction with the Windows Scheduler) is still a little raw.
Newest chipset drivers for b450:
Good article on and discussion of boost clocks and power plans:
https://www.techpowerup.com/forums/...-plan-for-ryzen-3000-zen-2-processors.260768/
You also most likely want
to turn off Precision Boost Overdrive, as the potential benefits don't seem to be worth the cost, even with excellent cooling.
For context, I'll give you a breakdown of my temperatures, before and after updating the software:
Old, stock cooler (Wraith Spire), PBO disabled:
- Idle temps = anywhere from mid-40s to high-50s degrees Celsius, with small bursts over 60. There will always be fluctuations because of the way Ryzen boosts.
- Prime95 Small FFTs Torture test: ~90 degrees Celcius after about 5 minutes (I was too chicken to keep going)
- Encoding a video in Handbrake - ~82 degrees
- Cinebench: ~81 degrees multicore, ~65 single core
New drivers, same cooler, PBO disabled:
- idle temps = mid-30s to high 40s.
- Prime95 Small FFTs Torture test: 82 degrees
- Encoding in Handbrake: 77 degrees
- Cinebench: 77 degrees multicore, 61 degrees single core
Performance is unchanged. The Cinebench score actually went up a little bit, and average boost clocks appear to be identical under various loads. I'm still going to upgrade my cooler, but the new temps are at least acceptable, whereas my first reaction was that the stock cooler is trash.
(There are also a number of people out there complaining about super high temps on Zen 2 with aftermarket coolers; I imagine this voltage/boost clock issue is the culprit, not the strength of their cooling solution.)