"Core Temp" - CPU Temperature monitoring tool for the K8 series

[davidk21770]

The diodes being read here are uncalibrated in AMD cores. AMD claims +/-14C without calibration and AFAIK, there is no calibration method.

 

[Rev]

Coretemp is reading ~2-3C higher than TAT. Very cool program though.

 

[RollingThunder]

The diodes being read here are uncalibrated in AMD cores. AMD claims +/-14C without calibration and AFAIK, there is no calibration method.

Dave,

Thanks, must have missed that in all the posts. So essentially it's useless for AMD CPUs. With that kind of tolerance range of 28 C from AMD I would be better to to rely on relative Everest or Speefan reads.

 

[ochungry]

RT, the speedfan would also show the CoreTemp highest reading.
Speedfan shows 3 temp reading for mine- the #1, and #2 readings are what everest or MBM5 shows for Core 1 &2, and the #3 reading is so called the diod reading same as CoreTemp reading. I take this utility w/ a grain of salt and just use it to monitor the highest core temp. The way I justify this utility is to use it as a safe measure and set CoreTemp reading as my highest limit when i stress test. Another word if I am priming, I observe this cortemp to know when to stop the test. I stop when coretemp reaches 57c, even if MBM5 or everest show 45-47c. You never know wich utility is closer to actual core temp. Add both, the high (coretemp) and the low (everest, MBM5 etc) and take the average of the 2, might be more meaningful.
BTE, those pic you sent me look awesome. Looks like you got the temp under control w/ that AMD heatpipe and extra fans.

 

[RollingThunder]

RT, the speedfan would also show the CoreTemp highest reading.
Speedfan shows 3 temp reading for mine- the #1, and #2 readings are what everest or MBM5 shows for Core 1 &2, and the #3 reading is so called the diod reading same as CoreTemp reading. I take this utility w/ a grain of salt and just use it to monitor the highest core temp. The way I justify this utility is to use it as a safe measure and set CoreTemp reading as my highest limit when i stress test. Another word if I am priming, I observe this cortemp to know when to stop the test. I stop when coretemp reaches 57c, even if MBM5 or everest show 45-47c. You never know wich utility is closer to actual core temp. Add both, the high (coretemp) and the low (everest, MBM5 etc) and take the average of the 2, might be more meaningful.
BTE, those pic you sent me look awesome. Looks like you got the temp under control w/ that AMD heatpipe and extra fans.

OC,

Essentially I was averaging when I was using Speedfan. It always read the same as Everest within 1 C degree. That was before going dual core. When I would turn on the machine after sitting overnight I would read the BIOS then go into Windows allowing for a couple degrees increase and see what it was after idling for a while.

I always run Everest in a small window while stressing to watch. It never goes over 52 C in Prime95 and right now it runs about 47-51 C. I'll idle at about 30-31 C this time of year so a 20 C rise is probably not unusual with a torture test.

I haven't even tried to control CPU temps yet it has been running that cool with ordinary use. Wonder how hot it will get when I give the vcore 115v AC current?

 

[davidk21770]

My understanding is that the BIOS reads yet a 3rd temp sensor and this is the sensor reported by all of the motherboard monitor programs that don't go out and read them directly (the only 2 that will do that that I know of right now are the latest version of everest and this program). (This 3rd sensor IS calibrated). Since the BIOS does not execute idle cycles, it'll read high and can not be compared to the value in windows. MBM5, SpeedFan, Everest, etc. should all read the same unless they are adding a scale or offset to the BIOS readings (calibration factors can be added in both programs -- the calibration added by MB manufacturers is not necessarily that precise). The 3 speed fan readings are not these diodes. They are most likely your CPU, your MB, and your chipset.

 

[RollingThunder]

My understanding is that the BIOS reads yet a 3rd temp sensor and this is the sensor reported by all of the motherboard monitor programs that don't go out and read them directly (the only 2 that will do that that I know of right now are the latest version of everest and this program). (This 3rd sensor IS calibrated). Since the BIOS does not execute idle cycles, it'll read high and can not be compared to the value in windows. MBM5, SpeedFan, Everest, etc. should all read the same unless they are adding a scale or offset to the BIOS readings (calibration factors can be added in both programs -- the calibration added by MB manufacturers is not necessarily that precise). The 3 speed fan readings are not these diodes. They are most likely your CPU, your MB, and your chipset.

Dave,

Some tinme ago in the Asrock forum we discussed and noticed the same. The BIOS always seem to read 2-3C higher than all the Windows programs and Speedfan, Everest and MBM5 all read the same within 1 C. Even with OCHungry's method of averaging I don't think I could use Core Temp if it shows that far off on my computer and with AMD's tolerance for temperature.

I don't know how a programmer could scale that much of a factor into a program for AMD CPUs. I did shoot my temps one time with a borrowed laser thermometer but of course with the heat sink in the way you can't get right to where you need it accurately.

 

[davidk21770]

I think that these diodes are a development effort for a future chip when they get the bugs ironed-out. Until then, the sensor reported by the BIOS works pretty well for me .

 

[The Coolest]

My article which discusses motherboard temperature sensors and DTS has been posted just now here:
http://www.overclockers.com/articles1378

 

[RollingThunder]

My article which discusses motherboard temperature sensors and DTS has been posted just now here:
http://www.overclockers.com/articles1378

Coolest,

Could you explain to a layman what the CPU power states designate? My above posts were not meant to be derogatory of your program. I realize that you can do only so much without specific and better calibration information from AMD. With the obvious incorrect reading I received on my AMD CPU I felt I could not use your program reliably. Intel is a much different story. RT

 

[The Coolest]

Sure.
A CPU Power state is referred to as C0 (normal), C1 (halt), and some CPUs have C2 and C3. Basically when a CPU is idle and is in C1, it uses less power than what it would use in C0, it skips clock cycles (like throttling) to conserve power.

 

[The Coolest]

I updated the first post with the article link and a "Digg it" link so to increase Core Temp popularity. so please click on it and digg it:
http://digg.com/hardware/Dual_Core_CPU_Temps_a_program_to_monitor_EACH_core

 

[RollingThunder]

I updated the first post with the article link and a "Digg it" link so to increase Core Temp popularity. so please click on it and digg it:
http://digg.com/hardware/Dual_Core_CPU_Temps_a_program_to_monitor_EACH_core

Coolest,

I'm trying your program again after reading your explanations once more.

Do I understand correctly that tcasemax, CPU power state and TDP will not register on an AMD CPU? In other words, these are either grayed out on mine or don't show at all (CPU Power State).

By the way it appears running Core Temp under load matches closely to Everest with temps. Core Temp now runs within ~2C of Everest (always a little higher). I only checked your Core Temp previously at idle.

 

[The Coolest]

TCM, TDP do not show on particular Rev E models, as they have undefined TCM/TDP profile.
And yes, it slipped my mind to add that the CPU power states are only shown on an Intel CPU.

 

[Cheator]

Is there any way to make this a sticky? I keep losing it

 

[watchthisspace]

Will there be a possibility for when you minimize it to Tray instead of having the little white window thing, it will show the temp, so you don't have to click on it to find out what it is??

 

[QuietIce]

Will there be a possibility for when you minimize it to Tray instead of having the little white window thing, it will show the temp, so you don't have to click on it to find out what it is??

IIRC - Mine will show the temp if you hover over the tray icon ...

 

[The Coolest]

I'm working on adding a feature to show temps in the systemtray.

 

[ochungry]

I'm working on adding a feature to show temps in the systemtray.

It Would be a good idea to have an option for user to calibrate the temp. another word, a way to increase or decrease the reading similar to MBM5. This way the user can adjust the reading according to some form of calibration. Those who remove the IHS and install a temp sensor on the core can adjust the CoreTemp according to the probe.

 

[RollingThunder]

I'm working on adding a feature to show temps in the systemtray.

Coolest,

Your program is beginning to make more sense the more I use it. It is consistently showing 2-3C higher during short Prime95 runs of an hour or two than the methods used by Everest or Speedfan for detection. I would much sooner error on the side of a higher reading than readings determined by less accurate methods.

Blowing an AC 115v 3450 rpm fan used to reduce my Everest and Speedfan temps. Now that doesn't happen because obviously you really can't get into the CPU core that way with the heat sink in place.

Personally I would like to see temps in the system tray shown permanently while the program is running rather than by a hover method.