Question Are average temps more important than max temps?

[ak2000spooky]

It's been a while since I checked temps and I noticed my CPU max temps are a little high. Last time I upgraded my cooling system was in 2023 (new parts, coolant, thermal paste, etc.) I used HWiNFO64 to log data starting 20 minutes into a Total War: Warhammer 3 game and logged it for about 50 minutes. Average temps looked good. I assume the max temps were for a split second and probably not worth any concern. My fan curves are not too aggressive, starting to increase once water temp hits 30C (4x 120mm intake fans kick up to 1000rpm once water temp gets to 50C, bottom 140mm intake fan gets up to 800rpm, rear 140mm exhaust fan gets up to 1000rpm, and pump up to 1800rpm). Any feedback on the average/max temps?

Current system:

• MB: MSI Z790 MAG Tomahawk WiFi
• CPU: i7-13700K (no OC; minor undervolted)
• GPU: 3080 VENTUS 3X 10G OC (minor OC)
• RAM: G.Skill Ripjaws S5 32GB (2 x 16GB) DDR5-6000
• SSD: Samsung 970 EVO Plus M.2 NVMe (1TB)
• SSD: Samsung 960 PRO M.2 NVMe (512GB)
• Case: Fractal Design Define 7 XL
• CPU Block: Corsair Hydro X Series XC7
• GPU Block: Corsair Hydro X Series XG7
• Rad: Corsair XR7 480mm (front)
• Pump: Corsair XD5 pump
• Fans: 4x Noctua NF-P12 120mm (intake push rad)
• Fans: 2x Noctua NF-A14 FLX 140mm (1 intake bottom, 1 exhaust rear)
• Tubing: 3/8", 1/2" soft tubing
• TP: Thermal Grizzly Kryonaut
• Fluid: EKWB CryoFuel Clear

https://imgur.com/a/qWXCfIK

 

[Lutfij]

I tend to look at all three, min, max and average across the period of time when the system is taxed. Also keeping an eye on how long it takes to get close to min after the system's taxing run is over.

Fans: 4x Noctua NF-P12 120mm (intake push rad)
You could extract a little more performance out of that radiator if you had a second set of fans as pull.

 

[dwd999]

It would depend on the length of time your motherboard designer/manufacturer decides should pass before there is a step up in cooling. But MSI doesn't seem address this in its manual.

For example, with ASUS motherboards for CPUs the bios defines the 1st step up as occurring after 2.1 seconds. Then 2.8, 3.6, 4.2, 5.0, 6.3, 8.5, 12 and 25 seconds. So momentary peaks of less than 2.1 seconds don't even cause the cpu fans to speed up at all. This is apparently acceptable under their design standards. But for case fans the 1st step up doesn't occur until a peak reaches 12 seconds.

You could probably question MSI support for guidance on how long a peak in temps should last before the fans should step up and maybe they'll answer.

 

[CountMike]

It's been a while since I checked temps and I noticed my CPU max temps are a little high. Last time I upgraded my cooling system was in 2023 (new parts, coolant, thermal paste, etc.) I used HWiNFO64 to log data starting 20 minutes into a Total War: Warhammer 3 game and logged it for about 50 minutes. Average temps looked good. I assume the max temps were for a split second and probably not worth any concern. My fan curves are not too aggressive, starting to increase once water temp hits 30C (4x 120mm intake fans kick up to 1000rpm once water temp gets to 50C, bottom 140mm intake fan gets up to 800rpm, rear 140mm exhaust fan gets up to 1000rpm, and pump up to 1800rpm). Any feedback on the average/max temps?

Current system:

• MB: MSI Z790 MAG Tomahawk WiFi
• CPU: i7-13700K (no OC; minor undervolted)
• GPU: 3080 VENTUS 3X 10G OC (minor OC)
• RAM: G.Skill Ripjaws S5 32GB (2 x 16GB) DDR5-6000
• SSD: Samsung 970 EVO Plus M.2 NVMe (1TB)
• SSD: Samsung 960 PRO M.2 NVMe (512GB)
• Case: Fractal Design Define 7 XL
• CPU Block: Corsair Hydro X Series XC7
• GPU Block: Corsair Hydro X Series XG7
• Rad: Corsair XR7 480mm (front)
• Pump: Corsair XD5 pump
• Fans: 4x Noctua NF-P12 120mm (intake push rad)
• Fans: 2x Noctua NF-A14 FLX 140mm (1 intake bottom, 1 exhaust rear)
• Tubing: 3/8", 1/2" soft tubing
• TP: Thermal Grizzly Kryonaut
• Fluid: EKWB CryoFuel Clear

https://imgur.com/a/qWXCfIK

Max temps may be more important if they go over Tjmax for particular CPU and hot spot for particular GPU because that's when they throttle down and lose performance and get closer to shutdown temps and that's dangerous.
Average temps are most used to determine cooler performance when bellow Tjmax and in normal use while running particular software/games as it constantly changes with usage.
Minimum temps are least important unless Delta between them and Tjmax is small which could indicate either bad settings (pimp and/or fans speeds) or incorrect mounting. Those are usually measured at CPU/GPU idle which could be very variable depending on what is running at that time. and could indicate too many background SW running. For Windows that would be about 1% of CPU and about 10% for GPU.
To get better data for troubleshooting, lower Polling period in monitoring software specially when using liquid cooling as it has longer delay between temperature raise of the source and radiator temps than air coolers. Depending on pump and amount of liquid that could be even several seconds during which source could be overheating/throttling and you wouldn't know about it.
With HWinfo for instance, default is 2000ms but I usually set it at 20ms.

 

[geofelt]

An Intel engineer quote.
"If you do not occasionally reach tjmax(100c.) you are losing some performance."

It is not the high temperature that might damage a processor but the voltage that caused that temperature.
If you are running well, and touching 100c. occasionally for a moment, I would not worry much about it.