Stuck DTS sensors on my QX9650

Clive Staples

Distinguished
Apr 11, 2008
22
0
18,510
0
Thanks for the excellent primer on temp (sticky at the top), CompuTronix.

Your post mentioned that processors with defective thermal sensors should be returned. If I'm understanding my "Test Sensors" results from RealTemp, at least two of the cores have stuck DTS. (e.g., 0 6 8 1 and similar results)

My question is, what exactly are the ramifications of the stuck sensors?

From the note in the sensor test utility, I gather that the test mainly aims to gague the sensors' ability to provide accurate temp readings at idle, but not under load. The note in the tool mentions that Intel doesn't recommend using the sensors to measure idle temp; they're designed for throttling and shutdown when things get too hot.

I ordered every component in my new system from newegg.com, including the CPU, and I have to say I've been very disappointed with newegg overall. Of my two Ratpor X HDs I received from newegg, one was a repack in an after-market anti-static bag, with a serial number that Western Digital said is invalid. (The box also had no SATA cable or any documentation/inserts, unlike a new one.) I returned it to them three weeks ago and still haven't gotten refunded yet, though it looks like that refund is finally imminent. I had other problems with my order, too, but I digress....

Now if I RMA my CPU I'm gonna have no system at all for at least 3 weeks (should have kept my old one a bit longer I guess); a prospect that doesn't appeal to me.

I never originally intended to OC anything when I placed the order, but as I've been learning more, the prospect of OCing seems more appealing.

Sorry for rambling... So, do I understand correctly that the stuck DTS mean it's difficult to accurately determine the Tjunction max for my particular CPU since I'll be unable to get an exact baseline reading for idle temps?

In practical, worst case terms, does having these stuck DTS mean:

A I shouldn't OC at all, period.
B I shouldn't OC very much, because I can't calculate the limits as accurately as needed
C I shouldn't use the CPU at all (even without overclocking), because it might really be running hotter than it should and I'd never know?

Under the circumstances, I think I'd rather keep the CPU and just not OC than send it back--IF the only consequence of a stuck DTS is that OC becomes riskier. But if the ramifications are greater than that, maybe I do need to bite the bullet and get it replaced.

Any advice would be appreciated, thanks. :)
 

CompuTronix

Splendid
Moderator
Clive Staples, most users are "stuck" on Core temperatures (Tjunction), while they tend to ignore CPU temperature (Tcase), which functions just fine. As I've gone to great lengths to explain in the Temp Guide, there are known relationships between Ambient, CPU and Core temperatures.

While Tjunction Max values still remain a closely guarded secret by Intel, and the proliferation of faulty DTS sensors should still remain unacceptable by users, it is not necessary to depend on Core temperatures to overclock. I have yet to encounter a "stuck" Case Thermal Diode (Tcase) circuit on any Core 2 Dual Core or Quad Core processor.

Intel depends on Tcase to monitor temperatures, and the DTS circuits for over-temp protection such as Throttle and Shutdown. Since CPU and Core temperatures scale very closely with one another, average Core temperature is easily interpolated from CPU temperature.

Regardless, keep in mind that since Tcase Max will be exceeded long before Tjunction Max (Shutdown) is reached, Tcase Max is always the limiting Thermal Specification. All that is needed to determine a very accurate CPU temperature is to follow Section 9 (Calibrations) in the Temp Guide.

I hope this helps to answer your questions.

Comp :sol:
 

Similar threads


ASK THE COMMUNITY