Crazy CPU temps

neeyik said:
The difference in reported CPU temperatures mostly comes down to the sampling time - the CPU package one is being reported directly by the CPU, and it probably updates the value every few milliseconds. This is to ensure the CPU can shut itself down quickly enough, if it overheats.

The motherboard's thermal sensor for the CPU will be sampled at a much slower rate, as all it's doing to checking the overall temperature of the chip that's plugged into it.

All modern CPU+motherboard combinations do this: for example, the system I'm currently gives these readings:

View attachment 86494

Note the difference in max temperatures between what the motherboard is reporting, compared to what the CPU is reporting. The faster sampling rate of the CPU has picked up the temp spike, whereas the slower motherboard hasn't.
Click to expand...
Are your fans REALLY running at 4k to 7k RPM???
 
Ha! No, the sensor doesn't like the AIO header connection. It's definitely not like this:

 
So I was going to install my new CPU cooler, but I didn't have the right brackets. So I ordered the brackets. But I wiped all the old thermal paste off, but I was an ***** because I forgot to check if I had any thermal paste lying around. Since I had to use my computer before the brackets came, so I got some thermal paste at the store. I put It on and re- assembled my PC. It helped lower the temps. Quite a bit, actually. They are still fluctuating, just not as fast, and the idle temp has dropped. I will still install the new cooler, mainly because it looks a lot cooler than the one I have, but also I don't want to have to deal with returning it. And, it will still keep my CPU cooler than it was, and I will be able to do strenuous things without having to worry about the temps.
 
I am probably going to do the reduction of voltage. Will dropping the voltage reduce the performance of the CPU?
 
Not directly so, but if the core voltage is reduced too much, then the CPU will become unstable at higher clock speeds. This is why you lower it in small steps and thoroughly test the system, for each one.
 
Ok, thanks! So I'll start out with 1.35 volts, then slowly increase if the system is unstable. What would be a sign of an unstable system? and what would be a good thing to thoroughly test it?
 
Signs of an unstable CPU = games crashing to desktop, BSODs (Blue Screen of Death), programs freezing, etc. A really unstable system might not even boot beyond, or even to, the BIOS phase of loading.

With regards to testing, start light to begin with: just use the PC as you would do normally. If all seems fine, then give it a big of nudge with something like Geekbench 5 or CPU-z.

If that's all okay, then up the stakes somewhat and run something more intensive, such as Cinebench 20. And if you really want to grind things to dust, try OCCT or Y-cruncher.

Edit: Yes, it would be worth the voltage tests - even if it only shaves off 5 degrees from the peak temperature.
 
Have a read of this article:

www.anandtech.com

Overclocking The AMD Ryzen APUs: Guide and Results

www.anandtech.com www.anandtech.com

Although it's about overclocking, it does talk about adjusting voltages, so you can see how it can be done. I think your options are restricted to BIOS and Ryzen Master, although somebody with more knowledge/experience of Ryzen APUs may be able to suggest otherwise.

Edit: AMD's guide for using Ryzen Master can be found here:

 
Ok, so I went into BIOS and adjusted the voltage to 1.35. It seems stable, no BSOD's no crashing apps, CPU bench and stress tests were successful. But, youre gonna think I'm crazy, but it seems like the temps are a bit hotter.
 
When I'm doing the same thing that I was doing, the temps are jumping to about the same place. My CPU Fan rpm is the same. ( I didn't know exactly what the temps were while running today because I didn't want to disable Riot Vanguard, but yesterday the CPU was under the same stress, and it was going about the same temp as the CPU is going now.
 
They may well be slightly higher, but it could be down to the fact that you've been using and testing the system a fair bit - so there will be some residual heat within the case, keeping the temps up.

You can also just be an unfortunate victim of the silicon lottery, and you just so happen to have a 2400G that performs within specifications, but right at the top of end of its thermal window.
 
As long as it hits the right clock speeds and doesn't exceed 95 degrees Celsius, then the CPU is operating within specifications. So it's not a failure by any means - you've just got a hot cookie :(
 
Ok well, that's 40 bucks for a new cooling system that I'll never get back :(. Well, at lest the new cooling system looks cooler.
 
Anyway, thanks so much for helping me out with this! It was very helpful! I guess that means I shouldn't use my CPU for anything tough like Folding at Home :(. I guess I've had it for 2 years now without a thermal throttling problem. Too bad it isn't something I could file a warranty claim for:(
 
Do you think I should turn my voltage back up? Because it didn't really do anything turning it down?
 
Sorry we couldn’t solve your problem, although a different cooler might help. You might as well set the voltage back to the default voltage.

One quick question: were you getting CPU usage and temperature spikes before you installed Valorant?
 
You know, that's a good question. I'm pretty sure I installed valorant before I started seeing these temp spikes.
 
You could fully uninstall Valorant and Vanguard , check things over again, just to confirm/provide piece of mind.
 
You must log in or register to reply here.

Similar threads

Shawn Knight
Delidding Intel's Core i9-14900K can drop CPU temperatures by up to 12°C
Replies
12
Views
799
bandit8623
B
Daniel Sims
See how an i9-13900K and RTX 4090 perform when it's -63F outside
Replies
22
Views
1K
TempleOrion
TempleOrion
noeljm
I7 6700k - High Temp issue
Replies
0
Views
4K
noeljm
noeljm

Latest posts

Back