I have been having a small issue of late with my CPU. I am running an E6600 OC'd to 3.0Ghz stock voltages with a Zalman 9500 on a P5B Deluxe. My temps at Idle are: Tcase = 32C / Tjunction 44/42 respectively. This is after some small temp corrections following CompuTronix's fantastic temp guide. Anyway, lately I will notice a temperature alarm and the Tcase will be reading 82C! Both of the Tjunction temps are still in the mid 40's.
The first time I saw this I nearly had heart failure. I whipped out my IR temp gun and the temp on the HSF heat pipes were reading low to mid 30's. This erroneous temp is being reported in both Speedfan and AsusProbe. A normal reboot corrects the problem and I might not see this for another couple of days. I am suspecting that the Tcase sensor is damaged. Any thoughts?
I believe I have 2 more months to RMA the CPU. Is it worth the hassle or would you consider this a minor inconvenience? I had planned to lap the heatsink and the CPU this weekend in an attempt to hit 3.6GHz OC stable on air, but now I am concerned and thinking that I should wait to fix my temp issue first.
Actually yes. I updated my BIOS in February when I built the system and things have run fine since. Looking at the ASUS site there is an update or two available. I will update tonight when I get home.
I am not certain that this is the issue based on the fact that my system has run without a hitch until lately. It has been just recently that I have been seeing the odd Tcase temp, and even then it is only every couple of days at most. I can pass Prime95 for 12+ hours small fft's without a problem. It seems a very random occurrence. Is it possible for the Tcase sensor to go bad?
I will let you know if I see the issue after the BIOS update.
...(B) Tcase is acquired on the CPU Die from the CPU Case Thermal Diode as an analog level, which is converted to a digital value by the super I/O chip on the motherboard. The digital value is Calibrated in BIOS and displayed by temperature software...
...(D) Tcase and Tjunction are both acquired from Thermal Diodes, however, analog to digital (A to D) conversions are executed by different devices in separate locations. BIOS Calibrations, factory DTS Calibrations and temperature software can all be erroneous"...
Tcase instability sources include:
(1) Case Thermal Diode
(2) Super I/O Chip
(4) OS Drivers
(5) Temperature Monitoring Utility
It is possible for BIOS thermal algorithms, which are processor model specific, to become corrupt, so it is one of several variables to eliminate. Also, if you don't mind jumping through a few hoops, you can upgrade SpeedFan 4.33 to Beta 4.34. Of course, except for cross referrencing temperatures, it is recommended that only one temperature monitoring utility be continuously run.
Bah...and I thought I had your guide memorized! I will update Bios tonight. I actually have SpeedFan 4.34 Beta running on my wife's machine. I will update that as well. I am keeping my fingers crossed that it is indeed a corrupt algorithm and not a bad Diode or Super I/O chip. Thanks for the help.
Bios is now updated and as it turns out one of the updates was an update to the temperature algorithm. I also updated SpeedFan to 4.34 Beta 34.
The updates forced me to re-calibrate my temps. I was surprised to see that I had to offset my Tcase by -14C to get to +1C ambient. This is a definite change from prior (-1C to get to +1C ambient). Anyway once the Tcase was calibrated the Tjunction temps were at +12 and +11C respectively from ambient. Verified the temp delta with the Prime95 stress test and all seems normal.
I will have to monitor the Tcase carefully to see if I get the erroneous spike to 82C but I am keeping my fingers crossed that the BIOS update did the trick. Hats off to you CompuTronix. Thanks for the assistance. I will let you know if the issue resurfaces.
That's good news. I would also suggest that APIC and ACPI support should both be enabled in BIOS, as well as PECI, if so equipped on your P5B Deluxe. This may correct your large Tcase offset, however, as there are dozens of LGA775 processor models contained in BIOS, OEM deficiencies are disappointingly common. It seems that they correct one set of problems and introduce others. Regardless of whatever offset is required for calibration, if Tcase is stable and scales properly with Tjunction, then it's hard to argue with success.