High idle temps on new Ryzen CPU (5900X) in old motherboard

Hardware CPU
,
1

Howdy folks,

I recently upgraded my system from a Ryzen 1700 to a Ryzen 5900X and am getting significantly higher temps on the new CPU with all other factors the same.

Lots of info incoming, so get ready…

I’ve had my Ryzen unRAID server going for about 6 and a half years now, and I decided it was time for an upgrade. I can’t afford to upgrade the whole platform though, so I decided to go for a new CPU that would be compatible with my existing motherboard and memory so I could upgrade those later.

So, the motherboard is a Gigabyte Aorus x370 Gaming K7. I updated it to the newest BIOS version in order to get compatibility with newer Ryzen CPUs. My old CPU was a Ryzen 1700. I’m using the same cooler on the new CPU as on my old one. A Corsair H100i I can provide other specs too if needed, just let me know if they’d be relevant.

The cooler is not 6 years old, I had two of them die within the first couple of years of the rigs life, and get warranty replaced. This one has been going strong for about 4 years, and I have no reason to think it’s going bad.

When the 1700 was in the system, it “idled” (as close as the system gets to idle with unRAID and lots of Dockers and VMs running, maybe around 10-15% usage) in the 40’s or maybe 50’s if the office was particularly warm. Now with the 5900X in the system, at “idle” it’s usually in the 70’s or sometimes high 60’s. This is true in unRAID and in Windows on baremetal (tested in windows with Open Hardware Monitor, and in unRAID with the system temp plugin).

My initial install of the cooler, I had much too much paste, and I thought that was probably all it was. I redid the thermal paste much lighter (but verified that it still spread over the whole IHS) and got the same results. I reseated it again just to be safe and still nothing has changed.

What’s even stranger still, is when I’m gaming (in an unRAID VM with PCI passthrough) or doing something else to load the CPU, the temps don’t go much higher. Into the 80’s, usually low 80’s but sometimes higher. But it has never throttled that I’m aware of or shutdown due to temperature.

As far as I know, my Corsair H100i is still working. If I unplug it from the fan header, the temperatures quickly spike until I plug it back in, and I can hear the pump running. I have all the fans set in the BIOS to 100% all the time to see if it helps and it doesn’t.

What might I be missing here? Why is this new CPU so much hotter? Or is that just how this CPU runs and I should just not worry since it doesn’t overheat in gaming or anything?

The only other thing that changed in my system is that I went from a GTX 1080 Hybrid with a water cooler and one fan (on the card itself) and one on the rad, to a RX 6750 XT with just dual cooling fans on the card and no water cooling. In the process, I no longer have an exhaust fan on the back of the case, but I can’t imagine that change would make a 30 degree delta…

The rad for the Corsair is mounted to exhaust out the top of the case, and I have two fans for intake at the front of the case, for reference. I dusted out the PC during the upgrade so if anything I figured it might run a bit cooler now.

Sorry for the huge info dump, I just want to provide as much info as I can so hopefully someone can help!

Thanks in advance!

2

I can appreciate all of the details in your post, but you are missing the one relevant data point for any thermal discussion; power draw!

What is your currently “idle” package draw with the 5900X?

You also may need to consider that the 5900X is the “worst” Zen3 processor in terms of efficiency. Maybe no where near as bad as Intel under load, but still a factor non-the-less.

1 Like
3

Hi @E-Wasted thanks for your reply! I do not have this handy, but I will get it.

I don’t think I can get this info through unRAID, so I will need to reboot the server to Windows on bare metal to get it, I’ll do that and try to get back to you ASAP.

Also thanks for the information about the efficiency of the 5900X. If it’s truly just not a terribly efficient chip, I’m fine with that. If it turns out that those temps are to be expected for my setup, I can live with that. It just seems like a huge jump from before, so I wanted to make sure I’m not missing something obvious.

I’ll get the power draw for you soon!

1 Like
4

Hokay, here’s a screenshot from OpenHardwareMonitor on baremetal Windows. The temps in Windows are more usually in the 60’s (due to lower “idle” usage I’m guessing?) but still higher than I’d expect to see.

CPU Package usage is between 50 and 60 watts usually.
The cores though are only using a small fraction of that, which seems odd to me, but I guess I’ve never paid much attention before, so I don’t know if that’s unusual or not to be honest. Is that normal for so much power to be used by the package with such a small percentage being the cores themselves?

Hopefully this info is helpful! Thanks again for the reply @E-Wasted!

OpenHardwareMonitor_L0rRa6LCJA
OpenHardwareMonitor_L0rRa6LCJA609×715 22.1 KB

5

I am admittedly very unfamiliar with AM4 & Zen3, but 15 watts on the cores seem totally normal to me for AMD.

What gets my attention though is the 60 watts reported by the package. If this reporting is accurate (again, not too familiar), that is basically the same as an i3/R3 running 100%, which is to say not an insignificant amount of heat production. In that case the 60C idle temps would absolutely make sense and be totally “normal” for the power draw.

However, one needs to keep in mind that Zen3 is a chiplet design and the R9 processors use two CCDs, or core complexes, plus everything required to connect the two plus the memory controller. This could very well be expected behavior in which your temps would also line up accordingly.

The only way to improve your temps are going to be either dropping the package power draw, whether through undervolting or other means, or to improve your thermal performance by increasing fan speeds or changing configurations.

2 Likes
6

OK, well that’s very good information to have.

To really push things, while I was in Windows on baremetal, I downloaded and ran Cinebench to try to really push the CPU temps. Keeping in mind again, that in Windows, the idle temps are somewhat lower, more in the 60’s, at the start of the test, it quickly hit 70 degrees. But then it took basically the entire length of the test, like 8 minutes, to heat up to ALMOST 80 degrees, and it never quite hit it. That was with all cores pegged for like 8 minutes straight, or longer.

So that seems to point to like you said, maybe it just idles pretty hot. Because under load, it certainly seems to be perfectly capable of keeping cool with the setup I have.

EDIT: Oh, and it was still turboing the entire time during the benchmark.

1 Like
7

Staying under 80C with a full load on an R9 sounds like a good thermal solution to me! :+1:

2 Likes
8

How are temps and W with ECO mode?

1-6 core load is always hotter than full core load. That’s because of the 2 CCDs. You can cool 100W better if it’s spread between two dies. 4-core boost are the highest temps I’ve seen so far. Everything <90°C is within spec.

1 Like
9

id watch your voltage using something like ryzen master. i switched to am4 and my 3600 would be at 1.4-5 volts for literally no reason, i was getting 45-90c temps until i told it to relax

10

I’m going to watch the voltage and see what’s going on there and if that’s related.

11

What were your frequencies like during the load test? 5000 series starts to thermal throttle at about 80 IIRC. If your freqs are good, I would say this is a stable system. If not, there’s more investigation to do.

12

OK so I have made a discovery. It seems to have to do with AMD Performance Boost.

I have not tested turning this feature off and checking temps and stuff in baremetal Windows yet, I’ll try to do that tonight. BUT in unRAID, I was looking through settings in the Tips and Tweaks plugin and for some reason, got the idea to disable the AMD Performance Boost option in there just to see what happened. Almost instantly, temperatures dropped by 20+ degrees. Turned it back on, back to 70+ degrees with a light load and not massively higher when under a heavy load just like I described before.

So I guess my question is, is that how Performance Boost is supposed to work? Should it really cause like a 20 degree delta when there is ANY load? I noticed that in unRAID, if I turned off EVERYTHING, meaning all my dockers and VMs and got the CPU usage down to a couple of percent, the temps went back to a more reasonable area in the 40’s or 50s. Still warm but I know Ryzen runs hot and I don’t have the beefiest cooler in the world. But it seems that as soon as there’s any load on the system, say 10-20%, it goes nuts trying to turbo itself up and the temps shoot through the roof.

Is this how it’s meant to work or is something wrong?