Ultra 6 Fan speed

Don't worry, shows 50 degree c is by design.

The below is all about CPU temperature reading:
1. Intel processors have two temperature monitoring mechanism, thermal diode and Digital Thermal Sensing (DTS)
2. Thermal diode is not accurate, its accuracy is prone to PCB layout, component variation, to use it, the system needs to be calibrate individually.
3. DTS reading is relative temperature compare to Tjunction. The Tjunction is maximum temperature the CPU can safely reach, beyond Tj, the CPU may be damaged. Linux coretemp driver translates relative temperature to human readable temperature, i.e. in Celtics degree.

We decided to use DTS for RNDU2/4/6, Pro2/4, the following must be taken care when DTS is used:
1. During a period time after boot, DTS reading may be shown as “Resource temporarily unavailable”
2. During a period time after boot, DTS reading may be very low, e.g. < 30C
3. When CPU is overheating, DTS reading may be shown as “Resource temporarily unavailable”, Also kernel log shows CPU speed throttle kicks in.

Because of above reasons, we decided to do following:
1. For any DTS reading less than 50C or “Resource temporarily unavailable”, display CPU temperature as 50C, this is per Intel’s guide line: Anything below 50C is OK and should be treated as reference only
2. CPU overheating shall never happen, guaranteed by hardware design, we stress CPU to its TDP under 40C room temperature. So there will be no overheating in promised operational range. Software does not need to handle overheating reading of “Resource temporarily unavailable”.
3. Intel CPU has two hardware mechanism to prevent overheating should we fail our hardware guarantee(which is very unlikely): CPU speed throttle and CPU thermal shutdown

Display 50C for DTS reading if CPU temperature < 50 degree C.

Could one maybe suggest to have the inaccurate CPU temperature reading as is, and in addition a DTS line desplaying either OK or "Warning CPU above 50C!" It seems rather stupid to delete valuable system info just because it is not accurate enough for thermal protection.

Also if you do not believe/use the thermal readings, how do you throttle the fan speeds?

I think evan is saying that the CPU temperature isn't used to throttle fan speeds unless it is > 50C

I don't think its useful to display incorrect temperatures. Instead I would say "CPU temperature: ok" when the temperature reading is < 50

StephenB wrote:
I think evan is saying that the CPU temperature isn't used to throttle fan speeds unless it is > 50C I don't think its useful to display incorrect temperatures. Instead I would say "CPU temperature: ok" when the temperature reading is < 50

Erm the thermal readings are not incorrect, they are not the most accurate true, but on other atom chipsets (i have observed) its usually within a few degrees. Sure it's not accurate enough for thermal protection, but it is enough to know whether your cpu is running at near 30 or near 40. If I for instance observe a sudden increase in temperature it might mean the unit needs to be cleaned, or some other event that needs checking out. It's a useful tool, and I think it's a bad idea to delete information, sure the 50C threshold read should be there as well, but as an addition not instead of the thermal readings.

But my original problem were the fan speeds so lets get back to that, if CPU temperature is not used to control fan speeds what is? Because it really is quite annoying that my RPM speeds has increased at least 100 without system or harddrive temperatures changing AND ambient temperature dropping. Something must control that fan :)

edit: if the Readynas does not have a thermal temperature meter like other chipsets then disregard the above (i assume that it has both thermal diode and the DTS thing)