I never said it uses the internal diode. But it uses an in socket sensor that I made sure was pressed right up the backside of the CPU. So it should be reading fairly accurate temps.Quote:
Originally posted by MrLavender
But the NF7-S doesn't read the internal diode does it? That's why I'm expecting the temps to be about the same as my KX7 using the in-socket sensor...
Now I tested this against the Epox 8RDA+ which has a sensor in embedded in the board, miles away from the back of the CPU, so how do you expect it to give accurate temps? It will be reading in socket air temps, NOT the temps from the CPU, so it's bound to be low, the socket isn't air tight!!! I can also say that by using my highly sensitive hand that when the Epox board says 50C, the CPU heatsink is way hotter than 50C! Yet my on the Abit board the temps feel about right.
Anyway, like I said, I used 2 sensors from my DigiDoc 5 and used a thermally controlled Fan to get an idea of CPU temps, I had all 3 sensors placed right next to the CPU core, topside. I found the Abit NF7-S was about bang on, usualy 1 to 2C higher. But the 8RDA+ was reading 10C too low. I just got annoyed hearing people say the 8RDA+ was accurate and the NF7-S wasn't.
Put it like this, if we did use internal diode temps they would be 10C~ higher than most in socket sensors, but the NF7-S is an exception due to the fact you can get the sensor to touch the CPU rather than just reading the air temps in the socket.
Now Abit have ruined this by reducing the temp readings in the BIOS because too many idiots complained.
Look what happened to Epox and their 8k3A, it read from the internal diode of the CPU and obviously read more accurate temps than any in socket sensor. But as they were higher, as expected, some idiots complained to Epox so much, and gave them so much hassle that they went back to inaccurate in socket sensors. Holst(A Epox tech) of AOAForums confirmed this.
Read This Thread on AMDMB