1
\$\begingroup\$

Every integrated circuit, like any other component, have a working temperature range. The minimum ranges are in the commercial grade (if I did dot forget something), while the military/aersopace grade had the wider one.

In what category falls an CPU, or GPU that we are using in our PCs? I see many people scared (mainly enthusiasts) when their PC is working at 60°C. Looking around, I found some indications on some AMD processors (I have one of these) in which AMD recommends to not reach temperatures above 73°C. Does it means that we are in a grade even lower than the commercial one? Or it is a safety measure?

Are those enthusiasts too much paranoid?

\$\endgroup\$
  • \$\begingroup\$ Generally, lower operating temperature (25°C vs 75°C) improves the life expectancy of an electronic part. \$\endgroup\$ – user2943160 Jul 7 '16 at 14:01
  • \$\begingroup\$ Often, the temperature that can be measured is not on the die, but on the package or the heat-sink, so allowances need to be made for the maximum differential between the two. I have an AMD CPU that is good to 100 degrees C, but that temperature sensor is embedded in the die. \$\endgroup\$ – Mark Jul 7 '16 at 23:22
  • \$\begingroup\$ Yeah agree, but I'm pretty sure that the heatsink does not have any sensor, usually, since it is bare piece of metal, and usually it is useless knowing its temperature. It could be on the package or on die. \$\endgroup\$ – thexeno Jul 10 '16 at 9:43
  • \$\begingroup\$ The traditional temperature ranges, 0-70°C, -40°-+85°C are mostly for components that don't dissipate much. For devices like CPUs and GPUs, the manufacturer provides a maximum junction temperature, for example 105°C, and some thermal resistance characteristics between the die, the PCB and the package. One must then design a proper radiator to guarantee that for the range of ambient temperatures you need, the chip will not exceed its maximum junction temperature. Many CPUs have also internal thermal protection, halting the CPU before it fries. \$\endgroup\$ – TEMLIB Aug 6 '16 at 14:21
2
\$\begingroup\$

The manufacturer's reccomendation is always definitive. Outside it, you may have problems; inside it you ought to be fine.

I believe that modern systems (AMD and Intel) will self-throttle in order to maintain safe temperatures. This isn't entirely reliable if your cooling system is inadequate, but it will stop it from self-destructing. Temperatures above 125C are definitely harmful for normal ICs.

(I have a laptop which will report temperatures up to 90C before going into emergency shutdown. The thermal throttling doesn't quite manage to keep it in acceptable ranges.)

\$\endgroup\$
  • \$\begingroup\$ Infamously, AMD did not use to self-throttle their CPUs, causing a lot of grief with underperforming heatsinks. Sadly, I can't find references to this now, my search results drown in all the noise. \$\endgroup\$ – pipe Aug 6 '16 at 15:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.