CPU and core temperature
Moderators, sorry for the Nubian topic ... :)Actually, I wanted to ask when the readings in Everest show the CPU temperature - 42 C and the CPU / core 1- 32 C, CPU / core 2 - 32 C, CPU / core - 32, CPU / core -32 C,
for example, when loading the processor: the CPU shows 58 C and the kernel is 48 C. The question is: what temperature should you look at the CPU or the rate of the nucleoli?
PS config in profile ...
ZabrO_o
As you can see, not at all.
And for those who have a monitoring curve in the bios / ancient Everest.
As you can see, not all.
And for those who have a monitoring curve in the BIOS / ancient Everest.
Uh, what's not for everyone?
Hmm ... Why then does not give the correct temperature ... Suppose Everest is normal. So bios? I recently reflashed the BIOS, to the newest one. there were some problems.
After a couple of minutes of reflection, I realized what was the matter. As we know, the crystal is protected by a metal shield and lies on the contact board. It all depends on the socket, the quality of the cooler installation, the placement of the CPU sensor, which, as I understand it, in the NEW CPUs is located on a monolithic board or a place close to it, and it is this placement that leads to a logical explanation. As it seems to me, both Intel and AMD place the sensor on a contact textolite board, (or why is it there that has much worse heat-conducting x-ki than the heat transfer to the side of the crystal-metal case-cooler), since there is no point in putting it under the kryshka, and what the observation of the temperature testifies to (if the sensor was under the kryshka, the CPU temperature would never exceed the temperature of the cores, even if the cooler was installed by the lamer). So imagine a processor installed on your mother with a perfectly installed high-quality cooler (like mine :) with a hair dryer ii x4 980 45nm socket AM3 (with legs) 1.485V, Zalman X10 Flex) which you will never have, since it has been discontinued, at idle CPU temperature 32, cores 29-30 at room 25C. We load the processor and see how the core temperature after 1 second increased by 10 C and the CPU temperature starts to rise slightly, and after 5 minutes of the test the temperatures converged at 50 degrees. We turn off the load and see how much faster the cores (crystal) are cooled almost instantly. We load the processor and see how the core temperature after 1 second increased by 10 C and the CPU temperature starts to rise slightly, and after 5 minutes of the test the temperatures converged at 50 degrees. We turn off the load and see how much faster the cores (crystal) are cooled almost instantly. We load the processor and see how the core temperature after 1 second increased by 10 C and the CPU temperature starts to rise slightly, and after 5 minutes of the test the temperatures converged at 50 degrees. We turn off the load and see how much faster the cores (crystal) are cooled almost instantly.
From here we conclude that the sensor is on a board that has poor thermal conductivity and the value of which depends a little on the motherboard and the socket, namely, the socket with legs give off heat, it seems to me it would be better.
Therefore, in a simple value of the CPU temperature, a couple of degrees higher than the values ​​of the cores, and in the load, the temperature of the cores is several minutes higher until the board on which the CPU thermal sensor warms up. Poor thermal conductivity is not only poor cooling, but also poor heating to whom it is not clear. The circuit on the thermistor and on-chip thermal diode is accurate and it makes no sense to talk about sensor glitches, although there are troubles with software. Maybe in old models, the thermal sensor was under the kryshka in the new ones, definitely not
alduindragonlord wrote:
I have a cpu temperature of 140 degrees
If in Fahrenheit, then it is normal.
