uGuru vs. CoreTemp

[EyeVahn]

I'm OCing a Q6600 B0 on an Abit IP35 Pro with stock Intel air. (I know, I know... I'll get a better cooling solution down the road )

Being mindful of temperatures, I've gotten it to 3.0 ghz (333*9)

but.....

uGuru (which is the soft. that comes with the IP35) and CoreTemp (which everyone is familiar with) diverge greatly in reporting my temps.

Somewhere along the lines of 15-20 degrees Celsius - uGuru reports lower temps, CoreTemp reports higher temps.

I've lowered my OC to a meager 2.8 (312*9 @ 1.275) and at full load with prime95 stressing all 4 cores and stock intel cooling uGuru tells me I'm @ 49 degrees and CoreTemp tells me I'm @ 65 degrees.

Which one sounds more truthful?

Thanks,
-Ivan-

 

[RGE]

I'm OCing a Q6600 B0 on an Abit IP35 Pro with stock Intel air. (I know, I know... I'll get a better cooling solution down the road )

Being mindful of temperatures, I've gotten it to 3.0 ghz (333*9)

but.....

uGuru (which is the soft. that comes with the IP35) and CoreTemp (which everyone is familiar with) diverge greatly in reporting my temps.

Somewhere along the lines of 15-20 degrees Celsius - uGuru reports lower temps, CoreTemp reports higher temps.

I've lowered my OC to a meager 2.8 (312*9 @ 1.275) and at full load with prime95 stressing all 4 cores and stock intel cooling uGuru tells me I'm @ 49 degrees and CoreTemp tells me I'm @ 65 degrees.

Which one sounds more truthful?

Thanks,
-Ivan-

nobody knows, there is a current disagreement whether B stepping quads have tjmax of 100 or 85. Coretemp assumes tjmax of 100, uGuru is assuming tjmax of 85, thus coretemp will read absolutes 15C higher. Maybe one is correct, maybe neither is correct. What does your cpu temp report on load, compared to the core temps at load? You can use speedfan latest beta, 4.34, to get cpu temp.

 

[jchunter]

Believe Coretemp. However, your temps are way too high for 2.8GHz. A good, well mounted heatsink with adequate airflow should result in a temperature difference (Tload -Tidle) of 13 - 18 deg.

 

[EyeVahn]

load vs. idle

Idle W/ uGuru = 26
Idle w/ CoreTemp = 38

49-26=23
65-38=27

so.... you're saying a 23ish - 27ish jump in temps from idle to load is too much?

Why do you think my diff. is so severe?

thanks,
-Ivan-

 

[JamesXP]

I don't think they're that bad.

I do belelive motherboard software grabs from sensors on the motherboard not from the DTS inside of the chip.

 

[jchunter]

The Intel's goofy push pins often don't install correctly, so pressure on the CPU chip could be uneven. Also, the chip or the heatsink could be not quite flat. Poor airflow in the case is also a possibility.

 

[RGE]

Believe Coretemp. However, your temps are way too high for 2.8GHz. A good, well mounted heatsink with adequate airflow should result in a temperature difference (Tload -Tidle) of 13 - 18 deg.

The reason his temps may be reading too high is because cortemp may be wrong in assuming a tjmax of 100.

absolute temps = tjmax (guessed at by software) - DTS. Since intel will not release tjmax, I would not assume coretemp is correct, especially given B stepping quads. coretemp assumes desktop have the same tjmax as mobile cpus, even though intel has said that is not correct. That was well known before anandtech article came out...but can read this, esp. last paragraph on that page.
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3251&p=4

 

[EyeVahn]

hmm...

Im thinking of pushing to a 3.0Ghz overclock on stock cooler....

I was dismayed by the CoreTemp readings... but what ur saying is that I should forget about them alltogether because of the Tj.max?

Using common sense, and putting temp. probes aside,

3.0 on stock intel for Q6600 G0 ... does it sound unreasonable?

Thanks again ya'll,
-Ivan-

 

[jchunter]

Im thinking of pushing to a 3.0Ghz overclock on stock cooler....

I was dismayed by the CoreTemp readings... but what ur saying is that I should forget about them alltogether because of the Tj.max?

Using common sense, and putting temp. probes aside,

3.0 on stock intel for Q6600 G0 ... does it sound unreasonable?

Thanks again ya'll,
-Ivan-

3.0GHz should be easy. 3.4GHz on a G0 is not difficult. 3.6Ghz requires more effort, but I'm doing it with a 30 deg (load - idle) deltaT

Edit: see OC graphs at http://www.ocforums.com/showpost.php?p=5579175&postcount=54

 

[RGE]

Im thinking of pushing to a 3.0Ghz overclock on stock cooler....

I was dismayed by the CoreTemp readings... but what ur saying is that I should forget about them alltogether because of the Tj.max?

Common sense and temp. probes aside,

3.0 on stock intel for Q6600 G0 ... does it sound unreasonable?

Thanks again ya'll,
-Ivan-

I saw first post and thought it was B stepping, actually you state BO, but you are saying GO?

The GO q6600 was measured by a Fluke to be tjmax of 95, so it is likely either 95 or 100 tjmax, 85 that uguru is using is not possible for G0 q6600. In which case coretemp is reading either accurate or 5C too high.

That being said the Q6600 GO Tcase specs is 71C
http://processorfinder.intel.com/details.aspx?sSpec=SLACR
And that is for tcase which is always going to be lower than core temps at load. So you could certainly go up to that point using coretemp and given the ? reading 5C too high, and the other at least 5C diff. between tcase and core, you are certainly within specs. So if coretemp says 71, your tcase might be approx 61 using prime 95 load, but thats just a guess.

Then it is just how long you want your cpu to last, as intel states "Higher temperatures result in earlier failure of the devices in the system. For every 10°C rise above the operating range means a halving of the mean time between failures." But I assume if you keep it less than 71 tcase it is likely to last the 3 year intel warranty time.

Personally, as long as I am 25C or so away from throttling, setting coretemp to read delta tjmax in options, I dont worry about temps.