I'm troubleshooting an issue on a Dell Precision T3500. As part of my troubleshooting I've decided to try running a stress test using Intel Burn Test software.

This machine is a stock configuration with 12GB of RAM and a Xeon W3670 processor (nothing overclocked).

When I run IBT using the standard mode, SpeedFan reports a processor temperature in excess of 80C. I've seen numbers as high as 90C but even at that temperature the machine does not become unstable or crash. However, it seems way too high. This processor has a TCase of 67.9C according to Intel's website. I'm guessing that means I'm in the danger zone any time I go over that temperature.

I've checked the cooling system and everything looks fine. I've even took out the heat sink and reinstalled it with new thermal compound. This did not appear to make the problem better or worse.

Is there a discrepancy somewhere here in the way temperatures are measured or displayed? I've also tried using HWMonitor from CPUID and it reports the same temperatures.

Should I just let the Standard Test go and disregard the temperature outputs?


I would say that's outside the spec if it's accurate. It may not get unstable but the CPU will slow itself down the hotter it gets. However, it is possible to get incorrect readings if the software doesn't recognize the CPU and is using the wrong offset. When I have had this problem I usually end up trying a whole bunch of different monitors to see if there are any differences. Core Temp is usually pretty accurate but like anything it's not always.

It could be that the heat spreader has come loose. Happens some times if a heat sink gets clamped on with unequal force on all sides.

99% of the time I have had these problems it's because the heatsink isn't attached correctly. Either it's cocked sideways with unequal pressure, too much thermal grease, or uneven thermal grease.

I wouldn't disregard it. Find the cause, even if it's just a software issue. Otherwise you could be looking at instability or eventual failure.


Once I had a similar problem with my lenovo D10. After some searching i found an explanation:

I dont know the specs of your particular cpu, but in general there is a problem that these temperature monitoring programs, have a wrong reference temperature for many cpu's.

The temperature sensor only gives out the difference to the allowed maximum temperature (T max, wich is different (and higher) than T case). The point is intel doesnt allways tell the T max, so the programmers of these kind of programms have to make a good guess, and some times they are off. Some programms even allow you to change the reference temperature. If you use the real temp program, (free) it shows exactly the difference measured, between Tmax and the current temp. So as long that difference is positive your ok, but of course, the larger the difference temperature the better. Especially because in those T3500's one of your HD's is quite close to the cpu heatsink.

the longer story of my similar question and answer, with corresponding refferences can be found here: http://forums.lenovo.com/t5/ThinkStation/D10-high-cpu-temperature-72-C/td-p/1436115

The most important reference was: http://www.intel.com.br/content/dam/www/public/us/en/documents/white-papers/cpu-monitoring-dts-peci-paper.pdf

EDIT: for the T3500 there exist 2 different heatsinks. A standard Aluminium one, and one with copper heatpipes. The aluminum one is for up to TDP of 80W, the other one for higher TDP's your cpu has 130W so please check if you have the right heatsink!

I was a little late for you, but hope this helps also others with similar questions.

Not the answer you're looking for? Browse other questions tagged or ask your own question.