The new forums will be named Coin Return (based on the most recent vote)! You can check on the status and timeline of the transition to the new forums here.
The Guiding Principles and New Rules document is now in effect.

"CPU" temp reading vs. core temp readings

GaslightGaslight Registered User regular
I just installed my new CM Hyper 212+ and I'm working on overclocking my old Phenom II X4 965, with the goal of getting to 3.8-4.0Ghz. I'm stress-testing with Prime95 and monitoring the temperatures with HWMonitor, Asus AI Suite, and Speedfan.

The problem is that most of these applications report one temperature reading listed as "CPU" and then separate (always higher) readings for the "core" or (supposedly) the individual processor cores. I am totally confused by which reading I should actually be paying attention to in order to see if my temps are safe. All I've gotten out of a good hour or so of Googling is that everybody agrees Speedfan is really confusing and possibly inaccurate for monitoring temps, but the other applications I'm trying aren't really proving any more informative. I have read several sources which indicate that the "CPU" reading is from a sensor in the chip socket on the motherboard while the "core" readings are actually the cores of the chip themselves. However, the problem is that while I have seen people quote maximum safe temperature ranges for my 965, I have no clue whether the temperatures they're quoting (and the ones AMD lists in its specs) are the socket temps or the actual core temps.

As a point of reference, currently while idling Asus AI Suite reports a "CPU" temp of 33 degrees C, which is the same as HWMonitor has in its "CPU" field. However, below that HWMonitor shows temperatures for each of the four cores ("Core #0," "Core #1," etc.) at 35-36 degrees C. When stress-testing, the "CPU" reading will climb to a maximum of about 50-52 degrees and then seem to stabilize, while the individual cores will read about 10 degrees higher than that. Which, if either, temperature do I actually need to concern myself with?


Gaslight on

Posts

  • GaslightGaslight Registered User regular
    edited December 2011
    Still sifting through Google, and found an overclock.net thread from a couple years ago where someone with the same CPU as me had a similar question re: what the quoted safe temperature specs from AMD actually referred to:

    http://www.overclock.net/t/663718/maximum-safe-phenom-ii-955-965-etc-temps-tcase-or-core-temps
    Dear Gareth,

    Correct, as long as you are within the voltage and temperature guidelines you are ok. The temperature that should be monitored is the overall CPU temperature, not the core temp.

    Best regards,
    Jesse
    AMD Global Customer Care

    So if this is correct than what matters is keeping the "CPU" reading (also apparently referred to sometimes as "Tcase" and which is a reading from the motherboard chip socket or the center of the cap on the chip) within the safe operating range. Anybody care to confirm or deny? If that's right, than my chip seems to be doing nicely at about 3.8-3.9Ghz. *Hopeful look, fingers crossed*

    Gaslight on
  • AtomBombAtomBomb Registered User regular
    I always thought the opposite, but I can't say I have any sources I could site. My feeling is that if a core is overheating, it doesn't matter that the overall CPU temperature is acceptable. As an extreme example, if I have 3 cores at 40° and 1 core at 100° my chip is at risk of failing, even if the processor overall is only 55°.

    I just got a 3DS XL. Add me! 2879-0925-7162
  • GaslightGaslight Registered User regular
    Well, common sense would tell me to back off long before I ever had a core at 100 degrees. :D 100 degrees is obviously way too hot. The issue is identifying the point that's just a little too hot.

Sign In or Register to comment.