CPU overheating during gaming sessions

DChuv

Honorable
Jan 9, 2014
26
0
10,530
Hi, guys.

I have been experiencing an issue with CPU overheating while playing modern, processor-demanding games.

I bought my desktop (AMD FX-8350, Asus M5A97, GeForce GTX970, 16 Gb Ram) about 20 months ago and for the first year or so had not had any issues with it, whatsoever. But eventually my computer started shutting down after an hour or two of playing CPU-demanding games. Installing some temperature monitoring software confirmed my suspicions: while, for the most part, the CPU temps stayed in the 50's and 60's (C), they kept spiking into the 70's and eventually even 80's. The problem kept exacerbating, and soon after I started having problems with running even older games.

Getting the dust out of the case, cleaning the heatsink, and changing thermal paste did resolve the issue for a while. The average temperatures went down, but the pattern remained the same: the temperatures stayed in the 10's and 20's (I actually saw them go down 10 C, which, I believe, is uncommon) while the computer was idle, ranged from upper 20's to lower 50's when I browsed or watched Netflix, and went up to 40's and 50's, spiking into 60's and 70's, when I was gaming.

Two or three months later, however, I noticed that the temperatures went back up. It is not as bad as it used to be before I cleaned the heatsink, but it's getting there. If a year ago I could run a game like Witcher 3 for hours without any overheating issues, now the CPU temperature starts spiking into the 70's after 10-15 minutes of gaming and into the 80's after 30-60 minutes (from what I can tell it depends primarily on the number of NPCs around the character).

Any advice that you could offer would be useful. Thank you!

Denis
 
Solution
Most times, if using decent paste, and it's applied properly, the pc will be obsolete long before the paste requires changing. AS 5 is different. Junk, imho, but different. I haven't touched my paste in over 2 years and unless I decide to mess with the cpu cooler, I won't need to.

AMD temps at idle are extremely misleading. It's physically impossible to mechanically cool something like a cpu to below ambient temps. This includes liquid coolers and full custom loops, since the fan blowing through a radiator is still a mechanical process. It takes a chemical process like liquid nitrogen (LN2) or a peltier setup to go below ambient temps. So if your area is 72°F, that's @ 22°C, add in case air and at best you'd be looking at realistic...
I'm not sure why you believe a CPU under load heating up is overheating, 80c is a perfectly reasonable temperature for that CPU under full load.

That said you could have a bit of lint or fibers mucking up your thermal interface.

Use a coffee filter and take as much care as possible to prevent foreign material from getting on the CPU.

Use a drop of thermal paste about the size of a grain of cooked rice, think brown rice not basmati.

That should get you back to ground zero to see what temps do.

A better cooler might not be a bad idea.

 

DChuv

Honorable
Jan 9, 2014
26
0
10,530


Thank you for your reply!

My cooling system has worked just fine for about a year, so I assume that it is decent.

The case has 1 x 80mm Side Blue LED Fan, 1 x 80mm Side Cooling Fan, 1 x 120mm Front Intake Fan, 1 x 120mm Rear Exhaust Fan.

The room, unfortunately, is not very well ventilated, but the back of the PC is about 35-40 cm away from the wall.
 

amtseung

Distinguished
FX8350 on an Asus M5A97 is a fire hazard. I speak from personal experience.

FX CPU's will begin to thermal throttle above 60~ish degrees C. 80 is technically not okay, nor is 70. I somehow doubt your CPU is having problems though. I think whatever program you're using to read temperatures is just reading it all wrong, which is very common when it comes to AMD.

Download and install AMD Overdrive, and use the temp monitor within overdrive to measure temps based on thermal limit. (This is why very few programs can read amd cpu temps properly; the voltage readings are just different.) It's always accurate as long as the CPU is under load. It'll give you a number of degrees C, telling you how many C you have until you hit the thermal limit.
 

DChuv

Honorable
Jan 9, 2014
26
0
10,530


Thank you for replying!

I am not an expert, obviously, but I believe that the TJMax for my processor is 90 C (at least that is what the monitoring software suggests). And eventually it does reach 90 C if I don't turn the application off. Afterwards, it simply shuts down.

As for your suggestions, I did clean it up very thoroughly, and as a result, the temperatures went down (but not by much) for a little while. But a month or two later they went back up. Considering that the system worked just fine for a whole year with only occasional minimal cleaning, I am not sure if dust is the real issue here.

I am considering buying a better cooler (currently, I am using a relatively cheap Thermaltake cooler). The problem is, as I mentioned, it worked just fine for about a year. Does the performance of a cooler deteriorate after a while?

Thank you.

DC
 

DChuv

Honorable
Jan 9, 2014
26
0
10,530


I was not aware of this. Thanks.
 

Karadjgne

Titan
Ambassador
Amd cpus are a nightmare as far as temps go. AMD overdrive is somewhat reliable, but mainly its due to the cpu itself not having reliable temp sensors on the cores. Top end for that cpu is actually 62°C. But it never gets read correctly, so what you'll normally expect is 72°C. Beyond that, you are playing a dangerous game, just how soon before the cpu burns out.

There's 3 important things about how to keep a cpu within acceptable temps. 1) use of a cooler designed to be effective to do the job. 2) sufficient airflow to allow the cooler to actually work effectively. 3) cleanliness.
1) AMD stock coolers are miserable. They really won't be effective at anything like high loading.
2) as long as those fans are actually working as intended, and getting a breeze through and out of the case, no issue.
3) pc's require cleaning. Every 3 months or so, more often if in a dusty environment. This has a definite affect on 1 and 2, the higher the dust, the higher the temps
 

DChuv

Honorable
Jan 9, 2014
26
0
10,530


I started using the software (CoreTemp) because my system would shut down after 30-60 minutes of gaming. If I don't run this software and thus remain unaware of my CPU temperatures, it still does the same.

Thank you for your suggestions. I will try AMD Overdrive.

 

DChuv

Honorable
Jan 9, 2014
26
0
10,530


Thanks for replying.

As I indicated above, I am using a relatively cheap Thermaltake cooler. I am considering buying a better one, but I am not sure if it will resolve the issue completely. The cooler that I have worked fine for quite a while, and if we are to believe the temperatures that CoreTemp is reporting, it still cools down the CPU to less than 20 C while idle. So, I do not know if changing the cooler would do much.

I clean the case regularly. About two or three months ago I changed the thermal paste and cleaned the heatsink. Most people I know do not have to repeat this operation every two months to prevent their PCs from shutting down under a high load. And for about a year I experienced no issues even though I didn't change the thermal paste once.
 

Karadjgne

Titan
Ambassador
Most times, if using decent paste, and it's applied properly, the pc will be obsolete long before the paste requires changing. AS 5 is different. Junk, imho, but different. I haven't touched my paste in over 2 years and unless I decide to mess with the cpu cooler, I won't need to.

AMD temps at idle are extremely misleading. It's physically impossible to mechanically cool something like a cpu to below ambient temps. This includes liquid coolers and full custom loops, since the fan blowing through a radiator is still a mechanical process. It takes a chemical process like liquid nitrogen (LN2) or a peltier setup to go below ambient temps. So if your area is 72°F, that's @ 22°C, add in case air and at best you'd be looking at realistic idle temps of @35-40°C. Not 10°C. This is why AMD Overdrive is preferred for AMD cpus, it'll give you a TjMax value, basically how much temp you have left. The closer your cpu temp gets to TjMax value of 0°, the hotter your cpu is running, the more important it becomes to remedy that situation.

On the flip side, windows 10 isn't windows 7. Win7 ran, and it did just that, very little was going on in the background at any time, so cpu usage was minimal at idle. Win 10 is a restless beast, it's always got something moving around in there, so true idle is not easy to attain. This means the cpu is almost constantly in use, even if not by much, which means you can expect idle (relative idle at best) temps to be slightly elevated vrs old win 7 or even win 8 temps. Anyone on an intel cpu can attest, used to see idle at 32-35°C, now it's a constant battle between 32 and 45°C spikes. At idle.

There's also the cpu fan. It's old. Contacts, brushes, rotors, bearings etc get worn, that fan has seen millions of rotations. When that happens, rpms become slightly unreliable, temps go up, the plastic blades can change pitch and warp, becoming noisy and less effective, end result is a decent little cooler just doesn't work quite the same any longer. Largest antagonist is heat. Heat from the cpu, case heat, even bearing heat.

Replace the cooler, it's time. Something capable of @180w or better TDP. CM hyper212 evo or X, equitable. For a minimum.
 
Solution

Karadjgne

Titan
Ambassador
The U9S will handle a 140w cpu. This would mean it'll handle your 125w cpu no problem under optimal conditions, or even use the supplied low noise adapter or maybe a decent OC. It's definitely a step up over the old TT, but overpriced. You can get a Cryorig H7 that'll do the same job, just as quietly (looks better too imho) for ½ the price. The U12S is a much better cooler, the D14 better still and barely cost a little more that the U9S. If your case has the space, the D14 has been the cooler to beat, for years.
OC is funky though. A small to medium OC can actually lower TDP on some cpus, stock voltages being set so high, lowering the voltages can reduce the heat output or at least maintain with a higher frequency. Just a matter of balances.

The best advice I can give is slap in the largest possible cooler that the case and your budget will allow, you can't overcool a cpu, but you can undercool one easily. The offshoot of that is a larger cooler has more capacity to absorb cpu heat, thereby the fans can run slower, which means a lot less noise, yet maintaining good temps and still having room to grow with OC or insufficient airflow.
 

DChuv

Honorable
Jan 9, 2014
26
0
10,530


Thank you. I installed a different cooler and the temperatures improved by 10-15 C. So, this issue is resolved.

Yet, as it often happens, a new problem arose promptly. Now all my games crash to desktop...

I opened a new thread to address it: http://www.tomshardware.com/answers/id-3213742/games-crash-desktop.html