CPU core is getting a bit heated under stress. - Hardware & Software

So I was just playing some DoTA 2, and I was just curious what my temperatures were at... I took a look, and while i'm playing I saw the cpu core hit 67 degrees celcius. Is this pretty hot? Could I have not used enough thermal paste? Maybe I should upgrade the stock fan?

67 is fine under load. [editline]14th October 2012[/editline] Also most stock cpu coolers already have thermal paste on them.

[QUOTE=taipan;38030103]67 is fine under load. [editline]14th October 2012[/editline] Also most stock cpu coolers already have thermal paste on them.[/QUOTE] Thanks for the reply. I removed the thermal paste from the stock cpu cooler and used my own.

My laptop used to get up to 100 C. 104 was the shut off temperature.

[QUOTE=false prophet;38065006]My laptop used to get up to 100 C. 104 was the shut off temperature.[/QUOTE] Yeah that means its time to open it up and dust out the heatsink from the side of the fan.

Unless it's a HP, then it's normal.

[QUOTE=taipan;38030103]67 is fine under load. [editline]14th October 2012[/editline] Also most stock cpu coolers already have thermal paste on them.[/QUOTE] Not really. While most modern Intel desktop CPUs have a TJunct of 72-100C, it doesn't mean you should run them near that temperature. AMD CPUs have always had low TJuncts between 62-68C before they start damaging themselves. If the CPU is going over 60C under any conditions, it means that the case doesn't have enough ventilation or the CPU cooler isn't mounted properly or is crap. Stock Intel coolers have been crap since the beginning of the P4 days and should never be used.

[QUOTE=bohb;38122329]Not really. While most modern Intel desktop CPUs have a TJunct of 72-100C, it doesn't mean you should run them near that temperature. AMD CPUs have always had low TJuncts between 62-68C before they start damaging themselves. If the CPU is going over 60C under any conditions, it means that the case doesn't have enough ventilation or the CPU cooler isn't mounted properly or is crap. Stock Intel coolers have been crap since the beginning of the P4 days and should never be used.[/QUOTE] Damage doesn't happen until at least 90C+. I had some kid in my computer servicing class get a wire stuck in his stock intel heatsink which blocked the fan, and he shut it down as soon as he realized what happened (after a few minutes of runtime)... he booted it back up a couple minutes later, and it was STILL at 70c and falling. I can't imagine how hot it must've been when he shut it off, but surely well over 100, probably around 120 or more. It actually seemed to run fine afterwards, but he kept getting disk corruption later in the lab, so it looks like there was some minor circuit damage.

[QUOTE=mblunk;38123129]Damage doesn't happen until at least 90C+. I had some kid in my computer servicing class get a wire stuck in his stock intel heatsink which blocked the fan, and he shut it down as soon as he realized what happened (after a few minutes of runtime)... he booted it back up a couple minutes later, and it was STILL at 70c and falling. I can't imagine how hot it must've been when he shut it off, but surely well over 100, probably around 120 or more. It actually seemed to run fine afterwards, but he kept getting disk corruption later in the lab, so it looks like there was some minor circuit damage.[/QUOTE] Thermal Junction is the temperature at which damage starts happening. If a CPU specifies a 71.9C TJunct, damage starts occuring at 71.9C, not 90C. Just because the CPU still works, does not mean that the CPU wasn't damaged. If you benchmark the same CPU before and after the CPU was overheated past the TJunct, you will notice that the benchmark scores will be worse than before. This was first documented publicly when Toms Hardware was benchmarking one of the last P4s to be released (3.73 GHz Extreme Edition.) On each successive run of the same benchmark test, the CPU would perform 1-2% worse compounded over every test. It turns out that the die was generating so much heat that couldn't escape the substrate that it was damaging itself progressively. All x86 CPUs are subject to this type of damage if run over the thermal junction, even just briefly.

[QUOTE=bohb;38134065]Thermal Junction is the temperature at which damage starts happening. If a CPU specifies a 71.9C TJunct, damage starts occuring at 71.9C, not 90C. Just because the CPU still works, does not mean that the CPU wasn't damaged. If you benchmark the same CPU before and after the CPU was overheated past the TJunct, you will notice that the benchmark scores will be worse than before. This was first documented publicly when Toms Hardware was benchmarking one of the last P4s to be released (3.73 GHz Extreme Edition.) On each successive run of the same benchmark test, the CPU would perform 1-2% worse compounded over every test. It turns out that the die was generating so much heat that couldn't escape the substrate that it was damaging itself progressively. All x86 CPUs are subject to this type of damage if run over the thermal junction, even just briefly.[/QUOTE] I thought TJMax was just the definition of the temp sensor's most accurate reading, and that readings below (and above) the value can be inaccurate (which is why some people have impossibly low idle temps, near ambient?) Can you give a source on that definition? Wouldn't corruption start to appear over time in the form of bad sectors, BSODs, etc due to data being corrupted on-die if any type of damage was really being done to the circuit? Because I have an old Q6600 build that has topped 100C on at least one occasion that I know of and it still reaches months of uptime on end to this day, over 5 years later.

TJMax on Intel CPUs is the definition of the maximum allowable heat tolerance before core(s) start employing emergency measures in an attempt to cool themselves down (lowering clock speeds, lowering voltage, inserting tons of NOPs into the pipeline.) The thermal sensor embedded in the CPU core is generally pretty accurate, though the thermal sensors on the motherboard are usually way off (due to being under the CPU and usually only getting ambient temperature readings.) Some motherboards have BIOS options to send tones through the PC Speaker or shut down upon overheating, but these options (if present) are usually disabled by default in the BIOS and must be turned on manually. Some people erroneously assume their machine will automatically shut down if the CPU overheats, which is false.