Quick q9450 temp question

[tricky trees]

Hi there,

Just after a quad owners opinion on weather a 7 degree difference between a lowest and highest core temp reading when idle is worth trying to correct by re applying thermal paste or if this is about normal?
I ask because those extra degrees on one core are seriously hindering hitting a higher overclock.
Thanks for any advice in advance,
Panface.

 

[iluvgillgill]

i would reseat the cooler if i were you.because 7C difference dont look like a faulty thermal sensor.its more likely its not been seated properly or just software problems.

 

[hsarc]

I have seen a 6 - 7 degree difference on my q9450. I think that is the way the processor is. I have reseated my Heat sink which fixed how hot it was running. I am at 3.2 Ghz and with a 100% load I will get to about 43 - 48 degrees celsius between the cores.

 

[iluvgillgill]

maybe that is the case then.since you got the processor itself im not gonna argue about it.play cool tricky tree.

 

[bpogdowz]

there a lot of stuck core sensors on 45nm procs and more dramatic at idle. really doesn't matter what the cores read at as long as the CPU case temp is below the Intel thermal spec.

 

[sailer]

My QX9650 has about a 10c difference between core 0 and the other three cores when its idling along. The reason is that core 0 is being used and the others aren't. So having one core run warm is normal. If I run Prime95, the cores even out to within 1c of each other. So before you get too concerned about the temp of one core being higher, run Prime95 and check what the temp spread is after 15 minutes of so. If one core is still substantially higher, then you need to reseat the heatsink. If the cores are nearly equal, then they are doing fine.

 

[tricky trees]

Nice one for the advice all of you. I'll give the prime 95 a run later and see if the cores even out. At the moment my idle temps range from 42 on core 3 to 49 on core 0.

 

[tricky trees]

Tried Prime 95 and it marginally brought cores 1 and 2 into line but core 0 was always 6 or 7 degrees up. Core 0 hit 63 degrees in Real Temp 2.60 whilst running Prime 95. One thing i did notice though was Core Temp Beta 0.94 had ALL my core temps quite a bit higher than Real Temp.
Probably an age old question but which one's more likely to be right, or should i just believe Core Temp is right because it's the higher reading and i'm safer ocing trusting the highest reading?
Thanks again.

 

[bpogdowz]

I think the Q9450 is accurate with TjMax 105. The programs I've seen use TjMax 95, 100, and 105. Intel has not released the information for which one their thermal spec is referring to.

 

[randomizer]

It really doesn't matter what the TJMax is. If the processor isn't throttling then it is still at a "safe" temperature.

 

[MoeDaKilla]

There is no way to accurately gauge the temps of any of the new 45nm cpus unless u have an infrared gun. From my experience I have found realtemp to be the best method for measuring temps.

 

[iluvgillgill]

well i would say the program that reports higher temperature is better.because that give you a bigger safe margin when oc.but i do suggest you reseat it and test it out.

 

[tricky trees]

I just read the Real Temp documentation at http://www.techpowerup.com/realtemp/docs.php and the programmer makes a very good case for why he wrote the program and why he doesn't trust core temp.
I'm going to go with the Real Temp readings because my core temp readings do seem too high for a 45nm processor and reset my heat sink.
Hopefully that will bring down my core 0 temp down and then i can play around some more.
Thanks for the advice everyone, always appreciated.

 

[bpogdowz]

well you 10 degrees over then again why not just enable TM1 or TM2. however even in Intel's tech document it says that TM1 and TM2 doesn't full prevent any prolonged use damage when exceeding thermal specs. my mobo reads this with a t junction of 110. not 100 not 105 110. why would a mobo maker use a 110 t junction to convince the user that their setup is thermally safe?

keep in mind this is 4 wolfdales and the readings you get from a 105 tjunction I think isn't out of the question.

Take a 6400+ AMD with a TDP of 125W which is dual core mildly over clocked to 3.3Ghz you're pushing the max thermal spec of 63-65 on air. Now take Q9450 TDP of 95W and you get the max thermal spec of 71-72 with a over clock 3.4Ghz on air.

What if the Q9450 actually has a tjunction of 110 as reported by my motherboard. Everyones shooting over the thermal spec and reducing lifespan. I don't know of a 110 temp program either but I trust the hardware over the software any time. We're all doomed

 

[jatt5]

Well my bios is reporting 22c and real temp 2.60 says its 32c idle and loaded at 46c on Q9450 using Thermalright Ultra 120 eXtreme cpu cooler. Bye

 

[randomizer]

I don't know of a 110 temp program either but I trust the hardware over the software any time.

Where on the hardware does it say it has a 110C TJMax? The BIOS is just software, and it isn't reading anything, it is just programmed to display a 110C TJMax when you put in that chip. Only Intel knows the actual TJMax, which could be 80C for all we know.

 

[randomizer]

Well my bios is reporting 22c and real temp 2.60 says its 32c idle and loaded at 46c on Q9450 using Thermalright Ultra 120 eXtreme cpu cooler. Bye

The BIOS is reading a different sensor and then based on some form of calibration reporting that 22C. Another BIOS version might report 28C or 16C due to different calibrations. Also the BIOS isn't actually idle, there is a small load on the CPU.

 

[jatt5]

OK