CPU temps – acceptable?

Topic

New Reply

I recently rebuilt my rig with a new CPU and MoBo. Prior CPU temps ware 35°-40° C, but now 45°-67°C. Is this cause for concern? Should I clean off the factory-supplied thermal paste and apply Arctic?

Thanks for your advice.

Zig

Reply | Quote

Replies

If you installed a newer gen CPU, you should probably refer to the manufacturer’s specs for operating temp range. There is a limit to what air cooling can accomplish with modern processors (and/or the resulting fan noise that you can tolerate).

Windows 10 Pro 22H2

Reply | Quote

Which temperature (core, Tjunction, x86-package, etc.) and at what load level?

 

Dell XPS 13/9310, i5-1135G7/16GB, KDE Neon 6.2
XPG Xenia 15, i7-9750H/32GB & GTX1660ti, Kubuntu 24.04
Acer Swift Go 14, i5-1335U/16GB, Kubuntu 24.04 (and Win 11)

Reply | Quote

Core temp = 45-76° C; no load (or light load) No overclocking (I’m not a gamer). AMD 5600X 0n B550 Aorus Pro V.2. Don’t know what Tjunction refers to or how to measure it, but Tjmax (hey, that would make a good name for a store) is 95° C. Currently running 60-61°C while reading this thread on Firefox.

Zig

Reply | Quote

On this pc, I’m running a AMD 5600x on a MSI MPG X570 Gaming Edge Wifi motherboard. Writing this reply, I’m running 31 degrees C, as measured by MSI Center. Although the same CPU, I would imagine that most everything else is different. Different case, fans, CPU cooler, and airflow. So not a direct comparison by any means.

Since you changed the motherboard, one of the first things that I would do would be to check the fan curves in the BIOS. It is possible that the case fans and/or CPU cooler fan(s) are turning too slowly.  Simple enough to check and fix.

1 user thanked author for this post.

satrow

Reply | Quote

60C with a light load and 80C under full load is OK, but it suggests you don’t have enough cooling, especially if you have a GPU working hard in the same box.

Is the fan the “standard” one provided by the manufacturer? If so it should be fine to keep the temps below max for the CPU as you don’t game.

The standard heat sink paste will be fine. Changing it may make a few degrees difference, but the heatsink size and fan are the things that matter.

cheers, Paul

Reply | Quote

Zig wrote:

Core temp = 45-76° C; no load (or light load) No overclocking (I’m not a gamer). AMD 5600X 0n B550 Aorus Pro V.2. Don’t know what Tjunction refers to or how to measure it, but Tjmax (hey, that would make a good name for a store) is 95° C. Currently running 60-61°C while reading this thread on Firefox.

Zig

That’s toasty. Acceptable… yes, in the sense that it is not over the maximum for that CPU (TjMax), but I would want to bring the temp down if it were my system. With a desktop PC (that has a reasonable cooler… even a stock one), you should see better than that with low/no load.

Have you confirmed that there is not some rogue process operating in the background, pushing the temps up? What kind of clock rate are you seeing while idle when you see these temps?

What kind of cooler is it? Is the fan spinning when you see the high temps? Some PCs/motherboards have a silent mode that will keep the fan off as long as possible to keep things quiet.

TjMax is the maximum acceptable temperature inside the chip, using a built-in temp sensor on the silicon die. This is the critical temperature that the CPU will throttle in order to avoid going over. Modern CPUs won’t allow themselves to go above their maximum temperature… they will throttle (slow down) to reduce temps to stay within the thermal limits. If there is a very bad cooling deficiency, so bad that throttling is not enough, they will turn off.

Ideally, you never want to hit the throttling point. It won’t hurt the CPU to reach it (though extended periods of time at that temp could shorten its life), but it does mean reduced performance (and probably more fan noise). On laptops, hitting the throttle point under heavy loads (whether from temperature directly or the related RAPL, or running average power limit) is common and not unexpected, since the limited space inside means that the cooler has to be pretty small. It’s normal and part of the design, but it would still be better if they didn’t have to thermal throttle.

 

Dell XPS 13/9310, i5-1135G7/16GB, KDE Neon 6.2
XPG Xenia 15, i7-9750H/32GB & GTX1660ti, Kubuntu 24.04
Acer Swift Go 14, i5-1335U/16GB, Kubuntu 24.04 (and Win 11)

Reply | Quote

If its an Intel chip you can test the setup:

https://www.intel.com/content/www/us/en/download/15951/intel-processor-diagnostic-tool.html

AMD weren’t quite as forthcoming last time I looked. Beyond checking the BIOS is up to date enough to correctly identify the chip installed (so the core voltage is right) if you have installed a NEW cooler on a clean chip, the (slightly domed..) heat spreader on the chip should have optimally spread that paste (which on the OEM ones at least is usually a form of Arctic, but in a waxy base so it doesn’t make such a mess- “it melts” on first use.)

We did have one guy who managed to get it wrong as we used a CPU cooler clip to capture the ATX2 power wire to ensure it stayed out of various coolers on one design, and he managed to trap a wire under the corner of the heatsink against the plastic surround. No damage resulted as the heatsink was in contact, just not fully “down”. The problem was detected in burn in testing so a demo of Passmark Pro, or whatever’s better now, might help you be sure.

Reply | Quote

Zig wrote:

Core temp = 45-76° C; no load (or light load) No overclocking (I’m not a gamer). AMD 5600X

Here’s an article that mentions hotter Ryzen 5 CPU temps are by design, although most thermal discussions seem to revolve around a full load, rather than idle. If it’s maxing out at idle, there is probably an airflow related cooling issue in the PC.

https://www.pcgamer.com/amd-views-ryzen-5000-cpu-temperatures-up-to-95c-as-typical-and-by-design/

“AMD is not freaking out over its latest generation Zen 3 processors hitting higher temperatures than what users might be accustomed to, or comfortable with, based on temp characteristics of prior generation CPUs. In certain instances, a Ryzen 5000 series CPU may run at 90C or even 95C, which is “typical and by design,” according to AMD.”

Windows 10 Pro 22H2

Reply | Quote

Ryzens do run hot.  I’m currently sitting at around 40C on my 3700X (with the supplied Wraith cooler) while typing this.  45-76 is a heck of an idle range, though.  I think the above advice to check your running processes is a good start.  What are you using to measure the temps?  Some software thermometers can add to the temp with their own processes.  I always saw AMD’s own Ryzen Master recommended for insignificant overhead, so that’s what I use.

Also, can’t hurt to check that the cooler is seated well and all your fans are working.  A little while back, I noticed my temps were high and discovered that one of the clips on my cooler had unlocked, and gravity had pulled it partly free from the CPU.  That led me to notice that the original thermal paste was also in poor shape.  I cleaned it off, applied Arctic Silver, locked the cooler back on, and everything went back to normal.

Reply | Quote

No obvious running processes, except CoreTemp (used to measure CPU temp) and HD Sentinel (used to measure disk temps), as well as GDrive. I’ve checked fan speeds & increased them a little.

When I installed the cooler, I inadvertently set it adhesive side down on my work table briefly (there used to be a small plastic sheet protecting the thermal paste before installation). Although there was no obvious contamination of the paste, I’m wondering if some micro-crud may have gotten in the way, thus my original comments on perhaps replacing the paste.

At the moment, I’ve just used the rig for ~20 mins. (light surfing), and it’s holding at 41°  C; I’ll see how it goes during the day.

And yes, I do realize that a new install takes a few days for the paste to soften, spread and work at full efficiency.

Thanks to all – will keep monitoring.

Zig

Reply | Quote

A few days later – now running at 40-50° C. Will continue to monitor.

Zig

Reply | Quote

Have you tried putting a heavy load on it, like Prime95, to see what the peak temps are?

Dell XPS 13/9310, i5-1135G7/16GB, KDE Neon 6.2
XPG Xenia 15, i7-9750H/32GB & GTX1660ti, Kubuntu 24.04
Acer Swift Go 14, i5-1335U/16GB, Kubuntu 24.04 (and Win 11)

Reply | Quote

I tried Prime95; the CPU cores  got up to 95°C in 3-4 minutes, so I shut the program down.

Zig

Reply | Quote

While Prime95 is commonly recommended for stress testing, it was actually “designed” to find new Mersenne prime numbers. The “torture test” included in the program (meant to ensure correct operation when searching for those numbers) places a HUGE load on a CPU so a lot of computer enthusiasts (especially those who overclock their systems) decided to use it to stress test their systems.

Thing is, it’s not unusual for a CPU to run near its Tjmax, or even shut down due to thermal overload, while running Prime95 (especially the newer generation of CPUs that are actually designed to run nearer their Tjmax) so it’s starting to fall out of favor.

Because it was so highly recommended, I tried it out back in 2018 when overclocking the i7-6700K in my old Asus tower and it crashed due to thermal overload after less than 15 min so I stopped using it.

When I overclocked the i7-9700K in my new Asus tower back in 2021, I used the “free” OCCT Personal and the “30-day trial” of Aida64 extreme and, after letting each one run for 24 hrs non-stop as a final stability test, achieved a rock-solid overclock of 4.9 GHz for the CPU and 4 GHz for the DDR4 memory and the core temps all stayed below 85°C (Tjmax is 100°C.)

Since you’re not attempting to overclock, I’d recommend you try one and/or both of those and let run it for at least an hour to determine what your stable high CPU core temps are.

3 users thanked author for this post.

Zig, Microfix, JohnW

Reply | Quote