Wrong temp with Intel E4300

[kogi]

[denied]

Yeah, for me too.....it shows 10C lower temp

[thedigital99]

I am also having the same problem.... E4300 on a 956P-DS3 motherboard.

Coretemp shows 17 idle, TAT Shows 36, Speedfan shows 13... I am inclined to believe the TAT reading however... I find it hard to believe that my temp is 37 idle @ stock. TAT also detects my E4300 as a Pentium M...

[The Coolest]

I don't really understand this either.
But from what I've seen TAT seems to usually read 15C over the others.
Basically TAT, for whatever reason, thinks that E4300 has a Tjunction of 100C. I don't know if that is right or wrong, as the program is not really up to date. CPUID of 6F2 could be accidently cosidered as an ES chip or something of that sort and therefore it displays calculates temperatures from 100C Tjunction.

I would really recommend you read these:

http://www.thecoolest.zerobrains.com/Co ... works.html
http://www.overclockers.com/articles1378/

This will give you a better understanding of how it works. There is no way to incorrectly read a digital sensor, each value has only one value. The only difference could occur when its substructed from a different value, thus giving mixed up results on different programs.

[fgw]

hi The Coolest,

unfortunately i have to confirm this!

changed to a e4300 today and have to recognize coretemp seems to display wrong temps! coretemp shows idle temps a few degrees below ambient temp! thats enough evidence for me to believe there is something wrong. core temperatures cant fall below ambient temps on an air cooled system!

tjunction is displayed as 85c. don't know if this is correct for an e4300 but i have no reason to believe its wrong!

tat is not working on my abit aw9d-max, so i cant crosscheck what tat would display. the abit utility coming with the motherboard displays cpu temps about 7 degrees higher as coretemp. the same reading is displayed by everest (7 gegrees above coretemp).
i even suspect the abit utility because it seems tho show too low load tems too!

so far i trusted coretemp over every other program.

may be you have a contact to intel, to verify dts readout registers and the register holding the ttc activation temp! anything must be messed up!

ps. would be happy if you confirm temp readout with coretemp is still correct for e4300. this would mean i get load temps in the low 40!!!!!!
too nice to be true!

anyway, GREAT PROGRAM!

[fgw]

The Coolest,

i'm watching this thread over on the Intel® Software Network for some time now. you might want to have a look at the last few posts of this thread as intel has finally released some info on this issue. not sure if it's of great help but check yourself ...

http://softwarecommunity.intel.com/isn/ ... x#30222546

i know you have been there in the past, but as i said, intel has updated the information in this thread and the thread is finally closed now!

[The Coolest]

Ok, I fixed the E4300 bug. Next version should show correct temps.

Right now there's one more thing left to add to Core Temp and then it will be released.

[fgw]

hi The Coolest,

any comments, what went wrong?

[The Coolest]

Well, my best guess is that the Tjmax (Tjunction in Core Temp) is supposed to be read as 100C on these processor no matter what the MSR which Core Temp reads reports.

[fgw]

Well, my best guess is that the Tjmax (Tjunction in Core Temp) is supposed to be read as 100C on these processor no matter what the MSR which Core Temp reads reports.

have you had a look at the Intel® Software Network forum i mentioned in one of my earlier posts? if i understand this correctly, the single bit which is checked for 85c or 100c tjmax is valid in mobile processors only! so as we are using a desktop processor, you might read the correct register and bit, but this bit might not hold any valid information! i'm not even sure if it is 85c or 100c for the e4300. intel might have introduced something else on the newer processors ...

in my case, i had a reading of about 50c to 55c in cortemp on the e4300. if you changed your tjmax from 85c to 100c the real temp might be around 70c which is a way high but far more realistic then 55c.

anyway, would suggest to have a look into the last posts of the thread at the Intel® Software Network forum. you will understand this far better than i do. probably there is a totally different approach to get the base temp from witch dts value needs to be subtracted in order to get correct coretemp.

by the way, as a suggestion, would it be possible to display the plain dts value in coretemp also. so even if i don't know the absolute temp my processor is running, i know at least how far i am apart from ttc activation!

[The Coolest]

Yes I have looked at it. And from what I understood that other way is not what you use to detect absolute Tjmax.
There's no way to detect absolute Tjmax.
I really don't know what to tell you right now. More digging needs to be done.

[fgw]

Yes I have looked at it. And from what I understood that other way is not what you use to detect absolute Tjmax.
There's no way to detect absolute Tjmax.
I really don't know what to tell you right now. More digging needs to be done.

you might be right! i will have a closer look into this during the next few days. get back to you if i find any conclusion...
anyway, thanks for your work so far! shame on intel for not giving away the documentation!