How Hot is Too Hot for PC Hardware?

[Gastec]

I believe these temperatures do indeed reduce the lifespan of CPUs, but not enough for you to notice. Plus, as both AMD and Intel do not want to give up profit margins, the two try to launch products always running close to the limit in order to extract the maximum possible performance from that silicon.

Fortunately undervolt exists and it is possible to run Modern CPUs at - 20-30° than stock, giving up 5% performance or less.

They should undervolt from factory, from that factory that still sells us what are basically overclocked Sandy Bridges.

 

[neeyik]

Great article! Bookmarked it for easy reference. Now according to that chart, the Tcase for a 13700k is only 72c?? Mine runs up to 75c while gaming, but averages around 65c. Is that something to worry about long term? Raptorlake just runs hotter than Alderlake, thought that was generally accepted. But they both have the same Tcase @ 72C?

Also, the chart has a typeo; it shows the 13700k as having 8P+8E &16T, but it has 24T.

Thanks for spotting the typo - all fixed now.

With regards to the Tcase values, Intel's ratings are somewhat complicated. They don't give an outright maximum limit, as such. Instead, they give a peak figure measured by test engineers, for a given power level and controlled ambient temperature of 40C, using Intel's own TTV (thermal test vehicle).

In the case of the 13700K, this value is 62C for 125W (see page 94 of this pdf) but system builders often use a higher figure, such as the one quoted in our chart, because they use a different thermal solution to Intel. Having a Tcase averaging around 65C, and occasionally hitting 75C, while gaming is absolutely fine, as long as your Tjunction is nicely under 100C.

You can find all the thermal figures you want for the different Core generations in Intel's technical documents (here) but in general, the maximum value for the Tcase, measured using its TTV, goes like this:

11th Gen (8 core 125W) -- Maximum TTV Tcase = 72C
12th Gen (8P + 8E 125W) -- Maximum TTV Tcase = 62C
13th Gen (8P + 16E 125W) -- Maximum TTC Tcase = 62C

Essentially, think of these values as targets for system builders to use, when designing completely cooling systems (heatsinks, fan, thermal materials, etc).

 

[Hodor]

Come on industry, start using something better than silicon. We need chips that can withstand temperatures up to 600 degrees centigrade. How will we create a colony on Venus if our CPUs start melting at 150 Celsius?

 

[Gastec]

Come on industry, start using something better than silicon. We need chips that can withstand temperatures up to 600 degrees centigrade. How will we create a colony on Venus if our CPUs start melting at 150 Celsius?

And they should also be salted and with a sprinkle of perejil

 

[ddferrari]

I'm a noise freak also. I de-shrouded the GPU and made a custom shroud to fit 92mm fans.
The original 75mm fans spinning at 3000RPM lasted only a day for me.
For me the 92mm fans now top at 1200-1300 RPM. Same power limit reduced to 73%.
The rest of the fans also reduced according to size:
frontal 2x200mm max RPM is 600
inside and back 140mm max RPM is 800
Cpu sink is 120mm max RPM is 1000
I have set all fans with custom curves both BIOS and Windows software

I have exactly two custom fan curves: one for gaming and one for everything else.

The "everything else" curve is whisper quiet, while the "gaming" curve maxes every fan in my case. It doesn't matter to me if it's loud because I discovered what is apparently a little known secret... over-the-ear noise cancelling headphones.

I personally can't see spending a bunch of money on high-end components, only to purposely gimp their performance. I get the max out of what I paid for, and both the CPU and GPU rarely hit 70C in any game.

 

[redgarl]

Problem? What problem? The only problem is that you believe everything an employee of large corporations says.

Engineers are just employees. If the CEO of Intel or AMD says that a CPU has to run at X temperature to achieve Y performance in order to reach Z profit margin, instead of making a bigger CPU that is more balanced in consumption and temperature, engineers are expected to comply with that directive. They don't hold the ultimate decision-making power in the company.

The higher the operating temperature of a silicon-based CPU, the more rapidly the silicon degrades. This degradation is caused by impurity diffusion and manufacturing defects such as thermal oxidation and doping. Although degradation is more rapid at higher temperatures, it can still occur at temperatures below 100°C, There are studies showing this.

There is a direct ratio between degradation and operating temperature. It turns out that CPUs are known to have a lifespan of decades, so even if the lifespan is reduced by 50% you still wouldn't complain because the product would have already become obsolete or failed for some other reason.

I am an electrical engineer, so I know what I am talking about.

Engineers are basing their specifications on science and quality assurance... you are basing yourself on your perception.

 

[redgarl]

No temperature is considered too hot nowadays for PC hardware, so why stop at the boiling point of water? Stride to reach 220ºC, that's a good, roasty temperature, even for pizza.

Many semiconductors can safely operate above 100C and they are present everywhere like in the auto industry. Let's not even dare talking about transport or space exploration...

All in the meantime, you guys are CONCERNED about your living room PC, in a dusty room, at load at 80C... and you are freaking out because AMD designed their PBO to max out at 95C in MT loads...

This is ridiculous... I am running my 7950x with a Noctua NH-D15 and I achieved 38000 at the same temps than an AIO after a 10 minutes average run. You guys know nothing.

 

[WhiteLeaff]

I am an electrical engineer, so I know what I am talking about.

Engineers are basing their specifications on science and quality assurance... you are basing yourself on your perception.

My arguments are based on actual academic research, you're based on arrogance and didn't bring up any arguments other than "Duh, I'm an engineer and blah blah, trust mee. You guys don't know anything"

If I were to count how many times these infallible companies and engineers made mistakes and caused losses, you would have to listen a lifetime.

 

[hel0sn8]

I went with the poorman method: open case option to keep my old hardware cool. no need for intake/exhaust fan. just need to dust off CPU/GPU fan every 3 months using an electric blower. so far so good.

And if you continue this way - you'll become a poorer man. Dust is the single greatest killer of PC components. By allowing dust to accumulate then blowing it off, you're actually driving the dust in to crevices etc that it would never have normally reached. Especially not compared to a standard setup where intake air filters reduce the amount of dust that can enter a case and thus reduce the amount of times dust will need to be cleaned up. Not clever.

 

[Gastec]

People don`t understand that those temperature limits are from engineering those chips. Engineers knows more about semiconductors than a bunch of hardware enthusiasts.

The worst are people mistaking heat generation with temperatures. Having a CPU at 95C while consuming 200W will generate less heat than a CPU running at 85C at 350W.

Temperatures are important for maintaining the integrity of the hardware. If it operates between the engineering limits defined, then the hardware will outlast your upgrading cycle.

People need to stop SPECULATING on things they don`t have a clue about. Thanks youtube for being part of this problem.

Oh my Lord Physics teacher, what an abomination you have written! Please read again the law of conservation of energy!

 

[Hodor]

And they should also be salted and with a sprinkle of perejil

That's a great idea to increase sales!

 

[MaxSmarties]

Many semiconductors can safely operate above 100C and they are present everywhere like in the auto industry. Let's not even dare talking about transport or space exploration...

All in the meantime, you guys are CONCERNED about your living room PC, in a dusty room, at load at 80C... and you are freaking out because AMD designed their PBO to max out at 95C in MT loads...

This is ridiculous... I am running my 7950x with a Noctua NH-D15 and I achieved 38000 at the same temps than an AIO after a 10 minutes average run. You guys know nothing.

quite the contrary, you seems to know very little about the matter...

 

[rsmith222]

'Nvidia GPUs: 80 - 90'

Nvidia said it intentionally used transistors that can handle high temperatures when it designed Fermi.

 

[mountains]

So I am currently configuring monitoring software like HWiNFO and others to watch this kind of stuff on a second monitor. I am curious what EXACT stats people feel are important to monitor using desktop gadget "things", and how they do it.

I am particularly interested in this line from the article:

"Excessively high voltages are more of a concern than heat..."

HWiNFO reports a lot of voltage stats. Is there one or two that would be smart to watch?

Thanks!

 

[Bionix]

I went with the poorman method: open case option to keep my old hardware cool. no need for intake/exhaust fan. just need to dust off CPU/GPU fan every 3 months using an electric blower. so far so good.

The problem with open cases is the components without temperature sensors are being sacrificed because they now have no active air flow.

If its really old its best to just cut a hole and fit extra fans even if they are on the wrong side of the case panel lol