Mainly in the interest of knowing more about my computer’s capabilities in order to improve FlightSim performance, I conducted a stress test on it per the instructions in [this video] by JayzTwoCents. I came away with more knowledge and confusion.
In the video JayZ describes the Cinebench program as having the ability to stress “any” system but that’s not what we see. At about 7 minutes he tests a Ryzen 5 5600X, the same CPU I have. What he shows is that the CPU quickly speeds up to the boost clock rate but the temperature NEVERS rises above 72-74 degrees and it continues to run at the boost clock rate. It never even approaches the 95 degrees limit where the CPU would self-throttle and the clock rate decrease. How then is this a stress test? More precisely, how does this tell us anything about whether a computer is overheating (because of insufficient cooling) and therefore slowing down when subject to high loads in MSFS when we’re trying to maximize performance? Wouldn’t a real stress test increase the demand on the CPU until it at least reached its max temperature and started throttling?
When I ran the test, my computer quickly reached 95 degrees and stayed there the entire 10 minutes by throttling down to 4,200MHz. But is it over stressed or not, since it’s still capable of running? Strangely however, although my CPU fan (on the stock Ryzen cooler) was also maxed out the entire time, all three case cooling fans hardly rose at all (even though on default program curves). This seems to imply my cooler didn’t transfer the CPU heat to the inside of the case as one would expect (wouldn’t we?) and therefore suggests I do indeed need a new CPU cooler. Or is something else going on?
I hope this is clear though it took me quit awhile to formulate this question. Thanks for any comments/suggestions.
Maybe try to re-apply the thermal paste and see what that does.
For 150 bucks and about 1/2 hour time I put an AIO (liquid cooler) on mine. Works really well.
1 Like
Your test show you that the cooler of your CPU is not sufficient. You should apply a liquid cooler.
When I bought my PC there was an air fan as cooler on my I7-10700 (unblocked power to 130W).
When I made the stress test the first time, the temperaure went up to 100° in 30s.
I mounted quickly a liquid cooler 240mm. Now in stress test my temp not go over 70° even I continue the test for 1h.
1 Like
Wouldn’t a real stress test increase the demand on the CPU until it at least reached its max temperature and started throttling?
imo yes
my computer quickly reached 95 degrees and stayed there the entire 10 minutes by throttling down to 4,200MHz. But is it over stressed or not, since it’s still capable of running?
it is over stressed, you will shorten the hardware’s lifespan. not necessarily true with your hardware, reference Mayhem6633’s article a few comments down
i don’t know about performance impact other than the obvious, if you can get it cooler then it can throttle at a higher clock
now i’m curious what my results are…
(and to respond to the guy above me - the only reason to choose a liquid cooler in 2022 is physical dimensions in a tight case or shiny looks)
Real overclocking is much more than just limiting heat but of course it still needs to be managed. It’s precisely why factory overclocks are tested in a laboratory and not someone’s hobby room and even then there’s no guarantee.
Cinebench 23 stabilized at 66’C CPU package temperature on my Ryzen 7 3700X as measured by HWMonitor.
No OC is applied.
Fractal Meshify S2
5x 140mm intake fans (three of which PWM)
1x 140mm outlet fan
This is with a $50 air cooler
The first thing to do to improve your thermals is check your fans. It’s totally possible you have all fans pointing in, or all fans pointing out. Both scenarios will prevent airflow over the CPU and increase temps.
Next step is to remove dust from your stock cooler’s radiator fins. If it is jammed with dust that alone is enough to raise temps.
Next step is to re-apply thermal paste (clean it 100% and ensure full die coverage - watch some vids if you’re not practiced)
(Possible next step: Add another case fan, either intake or outlet, depending on your case’s shape and the most likely dead spots where air is not moving)
(Possible next step: Redo your cable management - on one of my first systems the cabling was so bad it actually prevented airflow)
Final step is to upgrade cooler
1 Like
I have the 3800x and its supplied Wraith prism cooler is considerably better than the Aerocool tower I fitted at first install. Admittedly with mesh on all sides my PC is practically open case but I’ve never ceased to be amazed at how cool everything runs
Yeah AMD has a reputation for not-trash stock coolers. I do not think OP’s stock cooler alone is at fault for 95C temps.
3 Likes
I have an older 3600 with the same TDP. Cinebench took my CPU not over 66 degrees. I have a bequiet cooler for up to 180 Watts. One Fan for the cooler, one for the case. The fan speed up only a little bit.
For a 5600X watercooling is nice but not necessary.
If your computer quickly reached 95 degrees I would check the apply of the thermal paste.
A quick google search turned up this article with an inteview of an AMD guy. Here’s a quote from the article, " Yes. I want to be clear with everyone that AMD views temps up to 90C (5800X/5900X/5950X) and 95C (5600X) as typical and by design for full load conditions."
So the newer CPU’s have different design temperatures than past chips and you therefore can’t evaulate them according to old rules. With that said, you seem to be on the limit so I would try to lower it some just to be safe. But you don’t have to lower it a lot according to AMD. I would check my fan curves in the bios and make sure the fans are running at 100% when you get above 85C. In addition, make sure you have good airflow in the case. Maybe improve the case fan placement as well. You probably can get acceptable temperatures with the hardware you already own or by adding another fan or two to your case.
Another alternative is to optimize your PBO settings to make the CPU run better where this often happens when you undervolt the CPU. Undervolting with definitely lower those temperatures. For example, initial testing of my 5800x with stock settings I was getting temps around 90C. After optimizing the PBO settings (there’s youtube videos how to do this) I now normally run in the low 70’s while also getting improved performance.
1 Like
I have the AMD 7 3700X CPU, the stock cooler ( Wraith Prism ) would be way better than the Stealth 5600 cooler. But noise levels were too high for me using the Prism and also the ramping up & down was annoying and the temps were in the 75° to 85° or higher range, it looked pretty with all the flashing LEDs but that was it, So I think the “Stealth” would not be ideal for your gaming rig.
I purchased the Fuma 2 as it has more room around the base for memory sticks etc, and it matched liquid cooling for temps and won easily on price.
Your CPU matched my cpu at 65W
This is my running temps now and no noise at all.
1 Like
First, thanks for all these responses, more than I expected. A couple of points for clarity. My computer is less than a year old, checked a month ago and no dust build up anywhere. My memory isn’t great but I’m almost positive the cpu came with thermal paste already applied, I just bolted the stock cooler on. Also, regarding the case fans, I’ve checked them multiple times, both by inspecting markings on the fans and tracking air flow, and they are correctly oriented: two into the case from one side, two blowing out form the other side: one at the top and the other at the side side. Also, unless I’m missing something the fans appear to work properly: they do change speed occasional, but ONLY a hundred RPM or so… And I’ve tried custom fan curves but seen no difference. There doesn’t appear to be any build up of heat in the case and that appears to be why the fans don’t speed up. That said, sometimes when I’m flying I hear the computer making more fan noise than usual but I see no evidence it’s the case fans that have speed up.
What seems mysterious to me is why the heat being transferred from the cooler to the inside of the case is not sufficient to speed up the case fans so that they cool the cooler.
I’ll be reading the articles and get back her.
Moved to self-service, peripherals for better visibility.
CPU fan runs fine, quickly maxs out though.
This is a bit concerning to me. I’ve never seen a CPU come with paste pre-installed.
I have seen coolers come with paste on the cold plate, covered and protected with a plastic film you must remove before attaching the cooler.
Do you remember if this is the process you followed? If you’re unsure, it might be worth it to remove the cooler to check, then clean and reapply new paste to be positive.
1 Like
I think you may have a misunderstanding of how CPU’s act in terms of power draw & utilization, and what a CPU stress test is supposed to do.
Wouldn’t a real stress test increase the demand on the CPU until it at least reached its max temperature and started throttling?
The test just continuously performs specifically constructed operations that are designed to cause a worst-case scenario in terms of heat output. A stress test cant just keep increasing demand on a CPU until it reaches a temperature. The heat output of a CPU is proportional to the amount of power it’s drawing. The power draw is limited by the motherboard in order to prevent damage to the CPU and its supporting circuitry, such as the voltage regulation modules who tend to get pretty toasty under load. Some motherboards let you adjust these limits for extreme overclocking with special cooling systems.
That being said, Precision boost overdrive isn’t going to infinitely keep boosting the core clock (and power draw) up and up until it reaches 95c either. It’s programmed to have a max boost clock that it can’t exceed no matter the core temp of the CPU.
If the cooler can dissipate more heat during the test than the CPU can output while Precision Boost Overdrive is boosting to its max clock speed limit, you’ll get the situation that Jay has in his video. Sticking to the max boost clock with temps below the limit.
My guess would be you’re not getting good contact between the cooler and your CPU. Or the paste has some issues. One way to tell would be to feel the air that’s coming out the sides of the CPU cooler while the CPU is at 95c. Just be careful not to get your fingers in the fans, or touch the potentially very hot heatsink. The air wont hurt, but the heatsink might. If the CPU is at 95c and the air coming out the sides of the cooler is not at least quite warm, then the issue is likely bad thermal conduction between the CPU and the cooler.
If the air IS quite warm, then I would guess either hot air is being fed into the cooler, or the cooler just isn’t able to dissipate enough heat to keep your CPU below the 95c limit at max boost clock and you may want to upgrade it. Hope this helps.
1 Like
Thanks for the memories! The paste was on the cooler and when I removed the film, not knowing the paste was there, I slightly marred the paste on one edge of the CPU, i.e., there was a thumb print on maybe 10% of the cooler surface. I was a little worried about that but figured that when I married the two surfaces it should easily smooth out the very minor smudges. Unreasonable?
First, thanks for the detailed and useful descrption of how the test is supposed to work and how the system works.
Second, I am now convinced you are correct. I’ve repeatedly tested the air temp coming from the CPU cooler fan and it was NOT anywhere near 95 deg. C. (Boiling water!) Given what BigCow74 suggested above, it seems likely the problem IS with the paste and inadequate heat flow from the CPU to the air space in the case. Since I was thinking of getting a better cooler, with an eye towards upgrading my CPU when the prices drop, this seems like a good time to do that.
1 Like
Definitely get that sorted out first. But after that, you would be wise if you looked into optimizing the PBO settings and undervolting your CPU. That will make a big difference in your max temp.
Basically, by undervolting the CPU will still run at its max boost clock but it will be using less voltage to obtain that clock speed… thus it lowers your temperatures besides being able to hold that boost clock speed for a longer period of time. It’s a win-win situation.
Here’s the playlist of videos I watched to do it.
1 Like
