Jean Delvare wrote: >>>Almost everything is updated, so it may work for you. >>> >>> >>It does...sort of, temps are still off, but not by much (vs. BIOS) >> >> > >Someone on this list (I can't remember who) once explained that the >temperature as shown in the BIOS screen is usually higher than the one >of an idling Linux system, because the loop that waits for a user event >in the BIOS screen is poorly designed and leads to a high CPU load. >Well, that may depend on the BIOS designer, but that may be your case. > >You could think that comparing the values with another OS such as >Windows would lead to better results, but that's not necessarily true. I >remember my good old Pentium II 400 system had an idle temperature of 38 >deg C under Windows 98 and 32 deg C under Linux (as reported by the BIOS >right after reboot, so it isn't a matter of which software was used). It >really depends on how each OS wastes its time, so to say. > >Anyway, you most probably don't need a 1 degree accuracy. (BTW, most >people don't know that but most temperature sensors have an accuracy of >3 to 4 deg C, although they generally have a resolution of 1 deg C.) >What matters is not the "absolute" value. What matters is to know the >"normal value" to be able to detect overheating. > > > True. Say, I really appreciate you taking the time to respond, this is one of the main reasons I love the opensource community. That..and damn slick software. ;) The main reason why I'm trying to be very accurate with my temps is I am experimenting with a new water cooled system. I've got to machine up a nice flow meter and interface it with the fan output. That way an ALARM means no (low) flow...which means, well, your toast. Thanks again! M
