lm_sensors ACPI conflict and excessive Mainboard temperature

[Apheus]

Hello,

Last weekend I migrated my PC (gentoo and Windows XP installation) to a new mainboard and cpu. The mainboard is an EVGA nForce 750i SLI. I would like to be able to see the different temperature readings and maybe fan speeds. Software manipulation is not neccesary, I just want to see at least the temps.

lm_sensors detects a Winbond w83627dhg chip, but when it tries to load the according module w83627ehf, I get the well-known "conflicts with ACPI region..." message in dmesg (http://www.lm-sensors.org/wiki/FAQ/Chapter3#Mysensorshavestoppedworkinginkernel2.6.31).

The solution (kernel parameter "acpi_enforce_resources=lax") does not work - same message. I do not even know if this would be save, I never used this mainboard before.

I can get the module to work with the kernel parameter "acpi=off". Then the module reports 3 different mainboard temperatures (temp1, temp2, temp3) and 5 fan speeds. One of the temps is spooky: 90.0 °C.

I confirmed this with windows: SiSoft Sandra Lite reports MB temperature to be at or over 90°C. The only values I ever saw are 90.0°C, 92.0°C and 94.0°C. Maybe the value is doubled somehow? 45-47 °C would fit perfectly with the other two sensor readings. The biggest heatsink is hot, but not too hot to touch it, even for a long time.

=sys-kernel/gentoo-sources-2.6.34-r11
=sys-apps/lm_sensors-3.1.2

Two questions:

Is it save to disable ACPI with "acpi=off"? Or will it melt my hardware in the long run? Will fans still be regulated like they are now? The graphics card driver has an acpi use flag, how will this affect the self-regulation of the graphics card (GeForce GTX 460, =x11-drivers/nvidia-drivers-256.52)?

What can I do about the excessive temp reading? If it is true, something is wrong with the hardware. If it is wrong, could a bios upgrade solve this?

Thank you

[dmpogo]

Can you see the temperatures in your BIOS ? What does it show ?

[Apheus]

[dmpogo]

The problem with sensors is that you never know which are active unless manufacturer provided the exact specs how did it wire the chip on a particular board.
It often differs from board to board even for slight change in the board model number. Among them voltage is worst - almost impossible to guess without a detailed reverse engineering.

Overall, the usual way (and not too satisfying) way is just to ignore what seems sensless.

90C on motherboard can't happen. Only temperatures which go that high are perhaps grahics cores (it is too high even from CPU), but graphics cores are not usually read through add-on sensor chips. temp1-3 are probably CPU, something like NorthBridge and Chassis (some sensor on motherboard which monitors
ambient temperature). Look what sensor type is shown - in my experience CPU is typically thermal diode, while the other two - thermistor.

Modern CPU's also report temperature directly, try using coretemp module for Intels and k8temp (?) module for AMD's adn see what they give. They should be
few degrees higher than CPU temperature reported through w83627dhg chip.

I don't think (but who knows ?) BIOS update will help. Perhaps the correspondent pin on the chip is not connected to anything useful and just provides some fake
reading.

[roarinelk]

[Apheus]

Thank you for your tips, I will live without the ability to read the MB temp sensors for now. Any chance there will be an ACPI compatible module for the w83627dhg chip, like CONFIG_SENSORS_ATK0110 for Asus Mainboards?