micky wrote:


I havent' done much work yet but these programs disagree about the
temperatu

Core Temp 1.5 says my cores are each from 35 to 37C.
Anvir says my CPU is 40C, though that's the same temp it says my C:
drive is. And
Speccy says each of my cores is varying from 40 to 48C. Just getting
mail for a few seconds will raise the temp, it seems.

Quite a discrepancy, if you ask me. I looked other times too and
Core-temp is always 5^C (9^F!) lower than Speccy. Don't they use the
same sensor?

The digital temperature sensors produce a raw number which
is relative. You must know the Tmax, to convert the raw
measurement to a value humans can use. Let's say Tmax is 100C
for my CPU, and the measured digital value is "70".

100 - Trel_measured = 100 - 70 = 30C.

When two programs carry out the same math, they must use
the same value of Tmax. Nobody tells them what to use.
It doesn't appear there is a special function register
with the manifest value of 100 stored in it. Different
processor families have a different value of Tmax.
And then, CoreTemp and Anvir can arrive at a different
"number for humans", because of the way Intel makes it work.

The CoreTemp measurement, has a sensor-per-core. But you
will notice that the offset/gain errors on the channels
are not the same. The error is worst at low temps (the
readouts from two cores could differ by 7C). When the
chip is blazing hot, all the channels should read
very close to one another. The chip in this example,
isn't even remotely close to blazing hot - the heatsink
is too big for that (overkill).

https://s30.postimg.org/x3sxf96n5/prime95_jump.gif

If you scroll down this page, you can see the size
of the cooler I use. I only run it with the
center fan in place.

http://www.guru3d.com/articles-pages...-review,1.html

Paul