Which software to believe?

OzZie · 7 Mar 2009 at 10:49

As you can see below, the CPU temperatures vary quite a lot. I was wondering whether this has anything to do with me having speedstep enabled?

I'm assuming Coretemp is the one to trust, considering the other CPU temperature looks a little too low to be true in all honesty.

95thrifles · 7 Mar 2009 at 11:08

Yeah trust coretemp rather than anything supplied by mobo
Though on another point your cores shouldnt really be that far apart, try downloading realtemp and running the sensor test, make sure everything is ok

SirConfused · 7 Mar 2009 at 11:34

I don't know what kind of temps others get but those look pretty reasonable for a C2Q IMO. I had a Q6600 for about a year and a half and often found the temps across different cores varying more than that and with my recent upgrade to an i7 the difference across cores is even more pronounced.

If you are really concerned Ozzie then get a different app, if you just want an opinion then trust coretemp, those temps look fine.

95thrifles · 7 Mar 2009 at 11:39

Your cores shouldnt vary by more than 1 or 2 degrees, if they do then its possible you have a problem

BigBruiserAl · 7 Mar 2009 at 11:57

Not sure if its the same for a dual core but mine have had 8c diff between cores since it was brought over a year ago and still overlclocks great

wolvers · 7 Mar 2009 at 13:37

95thrifles said:
Your cores shouldnt vary by more than 1 or 2 degrees, if they do then its possible you have a problem

Nah, 3 degrees is alright. 5 or 6 isn't that abnormal either.

Be more relevant under full load anyway.

SirConfused · 7 Mar 2009 at 14:11

Have a look at this: 8 degrees difference under full load?? Should I be worrying yet?

reflux · 7 Mar 2009 at 14:15

95thrifles said:
Your cores shouldnt vary by more than 1 or 2 degrees, if they do then its possible you have a problem

Hey, my Q9450/X3350 had 9 degrees between the hottest and coldest cores, and that was after lapping. 3c is nothing to worry about.

SirConfused · 7 Mar 2009 at 14:17

reflux is that clock quoted in your sig using WC?

decto · 7 Mar 2009 at 14:33

Core temperatures can be quite different. If one core is doing work, even processing background tasks then it will be hotter than the others. Depending on the CPU there can be 10 deg C between cores... perfectly normal and a feature of the design.

The cores in any CPU are never perfectly matched vs each other so they will mostly run at differing temperatures for a given load and voltage. Then you have the mechanical instalation. variations in surface quality of the silicon, the metal heat spreader, paste etc.... and that's before you get to the external heatshink / fans and paste.

Temperatures are measured inside the silicon of each core so thermal transfer from this core to the external HSF makes a big difference.

AD

In the OP, the low CPU temp is probably a board mounted sensor for compatability with old CPU's which bears no resemblance to the internal temp of the CPU.

CurlyWhirly · 8 Mar 2009 at 01:28

decto said:
Core temperatures can be quite different. If one core is doing work, even processing background tasks then it will be hotter than the others. Depending on the CPU there can be 10 deg C between cores... perfectly normal and a feature of the design.

There's a difference of 16C with my Q9650 (idle temperatures) :eek:

Still it works okay so I'm not that bothered!

WJA96 · 8 Mar 2009 at 09:46

95thrifles said:
Your cores shouldnt vary by more than 1 or 2 degrees, if they do then its possible you have a problem

It's completely normal for temperatures to vary markedly in normal use. After all, if a programme is only using core 3 then why should the others heat up as much?

I think what you actually meant to post was that under full load, a marked difference in temperatures is an indication that something might not be right.

SirConfused · 8 Mar 2009 at 11:36

My quoted temp differences were under full load. It all seems to be working really well at the moment other than those temp differences. I did a video encoding run last night and didnt have any problems, prime tests stable for a good few hours (not had time for a full 8 yet) but the cores are still showing up a fairly big difference across the cores.