Jump to content

Huge difference between ITE and Core Temp @ Full Load

Recommended Posts

Well I guess I know why that I can't get past 2980mhz usually running 2980mhz dual Prime95 ITE reports temps at 54C @ 1.53 vcore under full load.

Now I am testing 2800mhz with orthos and ITE gives me 40c full load @1.400 vcore and Core Temp reads 49c and 51c Whoa.... thats a pretty dramatic difference I would say!

Share this post

Link to post
Share on other sites

My experience with both of my x2 3800+'s, is that you have to pop the IHS to bring the core temp down. My core temp with swiftech storm and a triple radiator went up to 60c when running at 2.8 ghz and 1.525V. After removing IHS it is now down to about 35-40c. I have two x2 3800+ and I have same experience with both of them.

Share this post

Link to post
Share on other sites

CoreTemp reads the actual cores' temperature from integrated registers inside the core itself, while ITE reads the CPU socket's sensor.


I usually find that CoreTemp reports ca 8C-10C higher than software that reads from ITE, under 100% load.

Share this post

Link to post
Share on other sites

core temps are read off the DTS (Digital Thermal Sensor) embedded in the core(s).

NextSensor, CoreTemp, Everest Ultimate 3.50, A674Info, Speedfan 1.40

polls the core/DTS at 1 sec intervals


cpu temps are read off the Thermal Diode (What we have been using for years).

MBM5, SG, NextSensor, Everest, Speedfan.

polls the diode at 5-10 seconds


They ARE two DIFFERENT sensors in two DIFFERENT locations.


Very often idle temps will be the same for all programs.

Really doesn't matter what program you are using as long as we know which one it is. Apples to apples comparison is what we are looking for.


Not all programs will run correctly on all boards.

Programs often misidentify or mislabel the sensors.

It's all software so you can't trust any of them 100%

Neither the DTS or Thermal Diode are "calibrated" at the the factory.

Share this post

Link to post
Share on other sites

While SMARTGUARDIAN does appear low when compared to "Core Temp" I still do not trust the results enough to say which is wrong.

"Core Temp" really depends on the DTS being calibrated correctly from the factory. AMD does not guaranty the DTS value and that is one reason we do not see it's use widespread.


I also think there are some programming issues beyond this. Linux 2.6.19 has also introduced the same method for getting the temps on K8 based processors. The k8temp.c file allows lm_sensors to use the same processor specific method to get the temperature readings.


When you compare the "how it works" section for "Core Temp" with the k8temp release information it is the same method.


Keep in mind that the "new method" is using a different part of the core(s) for the source of the measurement. We will thus see a some variance due to the location of the internal probes.


Core Temp - How it Works



Compare Linux k8temp doc




-------- ------ -------------

Idle 35C/31C 22C/30C 26C

Load 43C/40C 40C/45C 32C


Don't get me wrong... I like CoreTemp and k8temp too I just don't trust what it is doing.


All of the tests were done on the same system. The idle temperature difference on core 0 between CoreTemp and k8temp is 13 degrees celsius. This seems a bit high since the idle temperature difference on core 0 is only 1 degree!


At Idle I am also seeing 28.7 at the waterblock with an external sensors.

Anyway you look at it my temps are good but I can not tell you what they are at the core for sure.




Core Temp Beta 0.94: 35C/31C






Core Temp Beta 0.94: 43C/40C




Linux IDLE

k8temp: 22C/30C



Linux LOAD

k8temp: 40C/45C




Direct link to images






My System

* Monitor: Dell UltraSharp 2405FPW 24" LCD

* Motherboard: DFI Lanparty UT CFX3200-DR Socket 939 ATI CrossFire Radeon Xpress 3200 - BIOS 07/28/2006

* CPU: AMD Athlon 64 X2 4400+ 1GHz FSB Socket 939 Dual Core

* Memory: CORSAIR XMS 2GB (2 x 1GB) 184-Pin DDR SDRAM Unbuffered DDR 400 (PC 3200) TWINX2048-3200C2PRO

* Video: ATI All-in-Wonder X1800 XL 256MB PCI Express

* ATI HDTV Wonder Remote Control Edition PCI TV Tuner

* Sound: Creative Sound Blaster X-Fi Elite Pro Sound Card

* Sound: Realtek ALC882 8-channel HD Audio

* Ethernet: Dual Gigabit LAN - Marvell 88E8052 and Marvell 88E8053

* Hard Disk-1&2: Seagate Barracuda 7200.9 ST3300622AS 300GB 7200 RPM SATA 3.0Gb/s x 2 (total 600GB) RAID-0

* Hard Disk-3&4: Seagate Barracuda 7200.10 ST3320620AS 320GB 7200 RPM SATA 3.0Gb/s x 2 (total 640GB) RAID-0

* Hard Disk-5: Seagate Barracuda 7200.9 ST3160812A 160GB 7200 RPM IDE Ultra ATA100

* DVD-R/RW: SONY White IDE DVD Burner Model DRU810A

* DVD-R/RW: SONY Silver USB 2.0/i.LINK External DVD Burner Model DRX810UL

* CASE: Thermaltake Tai-Chi VB5001SNA Black/ Silver Computer Case with liquid cooling

* Thermaltake A2309 iCage 5.25" bay (2 units)

* LCD-TEMP: AeroCool GateWatch-SV Silver GateWatch with LCD display

* PSU: PC Power & Cooling 510 SLI-PFC 510W

* UPS: APC Back-UPS XS 1000 UPS, 1000 VA, 600 Watts

* OS: Microsoft Windows XP Media Center 2005 Rollup 2

* OS: Fedora Core 6 - x86_64

* Speakers: Klipsch ProMedia Ultra 5.1

Share this post

Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in

Sign In Now
  • Create New...