What temperature utility is best?

ChrisLX200

ChrisLX200

ChrisLX200

Associate
Joined
23 Jun 2004
Posts
2,470
Location
Macclesfield
I'm having some difficulty figuring out what the real CPU temps are. I've run 3 different utilities at the same time (Asus PC-Probe, Real Temp 2.90, and SpeedFan 4.37). They give different results - and by a wide margin. I set speed to stock and ran seti@home to max out the cores for an hour, then took a snapshot of the data (below). The core temps in SpeedFan are about 15deg different from those in RealTemp. The single 'CPU' temperature offered by PC-Probe is similar to the CPU figure in SpeedFan. It would be nice to know which is the critical number to watch (and who knows what the 12V rail really is? - choice of 11.56 or 12.31)

temps.jpg
 
The only ones that are truly accurate are the distance to Tjmax in Realtemp. But for convenience sake take the Realtemp Temperature which uses the correct Tjmax of 100C.
 
Things change, they all use the same data from the cpu. All of them subtract the DTS values from the Tjmax. If the wrong Tjmax is used then you get the wrong temps. Intel recently released all the C2D Tjmax values, so most temp apps should agree for older cpus. Realtemp 2.90 is currently using the Tjmax correctly for Ci7s, the writer Unclewebb currently does a good job and keeps up to date with realeases. But that could change.

http://www.tomshardware.co.uk/intel-dts-specs,news-29460.html
 
Last edited:
I've always trusted Everest. I've compared it against the TJMax in Realtemp as well as all the other popular temperature software and it gives the same readings, so it seems pretty accurate.

It also gives a ton of other info about your PC so it's a nice all-in-one program.
 
I find that putting my tongue on the chip is a fairly accurate way of measuring temps - instant recoil = too hot.

p.s. Realtemp 2.90 for Core i7's
 
chuffnut said:
I get the same temps in Coretemp and speedfan , but Realtemp reports the temps 5c below the other two.
Click to expand...

Make sure you have the latest version 2.90. Take the value for the temperature and the value for delta to Tjmax. Added together they should always be the same as the Tjmax as published by Intel, below. Allegedly...

 
I've had a Q6600 (running at 3.2GHz atm) for over a year now and I've only ever used coretemp to monitor the temps on it because I read that was the most reliable.

I've always wondered why my temperatures were ~10 degrees higher than everyone else at idle.

After reading this thread I've altered the TjMax value in coretemp from 100 (default) down to 90. Now the temperature is 10 degrees less.

Just taking the temperature from core 0...

Core Temperature: 30
Delta to TjMax: 60

It's a bit chilly in my room atm @ 16 degrees

Both coretemp and realtemp agree on everything when I set TjMax to 90 degrees on both of them. Is this an accurate reading now?
 
It's tricky finding the TJMax for Core i7 as Intel don't specify it (why?), anyway the only independant reference I could find suggested it's 95deg - so that's what I've set it for.
 
Its 100C on a 920. Tjmax is a little different on Ci7s, it can be read directly from MSR.

http://www.xtremesystems.org/forums/showthread.php?t=205779&page=46
http://www.xtremesystems.org/forums/showpost.php?p=3445590&postcount=317
http://www.xtremesystems.org/forums/showpost.php?p=3501719&postcount=953

Easy enough to check if you have water cooling and a loop sensor. 22C ambient, my loop temp is 26/27C at idle, speedstep enable etc, so as little power used as possible. Realtemp shows 30C idle with 100C Tjmax. With 95C Tjmax my Realtemp idle is below my actual water temp, obviously impossible.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom