Sign in with
Sign up | Sign in
Your question

Intel Q9450 vs Q6600 Idle Temperatures

Last response: in CPUs
Share
a b à CPUs
October 3, 2010 2:20:43 AM

Specs: CM Mid-Tower w/ 2 fans
HSF: CM N520
Mobo: Biostar TP43D-A7

I originally had a E5300 OCed to 3.6GHz, and it idled at 30'C, and maxed out in the upper 40s'C.

Then I got a Q6600, minor OC to 2.8Ghz, which idled at ~31'C and maxed out at 55'C.

Recently I acquired a Q9450, and have it at stock 2.66GHz. It idles at 40'C and maxes out at around 55'C as well.

Doesn't the 45nm Q9x00 have a lower TDP (10watts lower) than the 65nm Q6600? Why is it idling 10'C hotter than the Q6600?

I used the same thermal paste and the same application method...did I botch up somewhere?

Is the reading correct? (HW-Monitor)
a b à CPUs
October 3, 2010 2:38:14 AM

CPUID HWMonitor is best, and I haven't heard of any specific problem with that cpu. I'll tell you as you likely have been told before, try Coretemp and Realtemp - probably yield the same.

You could check your voltages in BIOS (vs the Q6600) and see if the idle temp difference is related to voltage differences at idle. You could check and see if all the power saving settings are the same as you used before.

But if the max temp under full load for that quad is 55C, I wouldn't be concerned.
m
0
l
a b à CPUs
October 3, 2010 3:07:17 AM

My Q6600 voltages were around 1.15v idle, 1.25v load.
My Q9450 is around 1.12v idle, 1.21v load.

All my power settings are the same, and stepspeed/C1/etc are turned on. The Q9450 downclocks as normal just like the Q6600.

My max temps aren't really a concern...but for me, it's just really weird that the 45mn 95w TDP Q9450 is 10'C hotter on idle than the 65nm 105w TDP Q6600.
m
0
l
Related resources
a b à CPUs
October 3, 2010 3:20:28 AM

Yup. Well, something in there is working harder at idle - and perhaps slightly harder at load - than logic would predict.

Be interesting to see what happens as you OC it. Relatively speaking, its hot at load, warm at load, might be cool by the time it gets to OC levels.

I doubt it, but one can dream lol.
m
0
l
a b à CPUs
October 3, 2010 3:23:14 AM

Twoboxer said:
Yup. Well, something in there is working harder at idle - and perhaps slightly harder at load - than logic would predict.

Be interesting to see what happens as you OC it. Relatively speaking, its hot at load, warm at load, might be cool by the time it gets to OC levels.

I doubt it, but one can dream lol.


Weird...

I asked the same question on overclock.net, and some people said I was just unlucky with a "warm" cpu - basically the same as a lower binned CPU but just warmer than the average median temperature of the chips.


Another question, why is the 3.0GHz Q9650 TDP 95W, but the QX9650 TDP 130W?

The unlocked multiplier generates heat?
m
0
l
a b à CPUs
October 3, 2010 3:35:30 AM

Dunno. Maybe they expect the multiplier to be used.
m
0
l
a b à CPUs
October 3, 2010 6:09:20 AM

Bluescreendeath said:
Doesn't the 45nm Q9x00 have a lower TDP (10watts lower) than the 65nm Q6600? Why is it idling 10'C hotter than the Q6600?


How do you know it is? You're putting faith in sensors that are designed to trip thermal throttling and shut down, not report accurate core temps. Not only that, but you're comparing the reasonably good sensors used in 65nm Core 2s with the wildly unpredictable sensors used in 45nm Core 2s. The sensors may simply be bottoming out at 40C, and therefore wouldn't report any lower even if you put it on water (aka stuck sensors).
m
0
l
a b à CPUs
October 3, 2010 6:41:16 AM

randomizer said:
How do you know it is? You're putting faith in sensors that are designed to trip thermal throttling and shut down, not report accurate core temps. Not only that, but you're comparing the reasonably good sensors used in 65nm Core 2s with the wildly unpredictable sensors used in 45nm Core 2s. The sensors may simply be bottoming out at 40C, and therefore wouldn't report any lower even if you put it on water (aka stuck sensors).


So the 65nm C2Q sensors are better than the 45nm C2Q sensors? I didn't know that...

As for stuck senors, well, the temps for the Q9450 actually showed a min temps of around the mid-high 30s'C when I resume from sleep mode (cooler from bootup I presume). It slowly increases to around 40'C and stays there while idle.

This is the same problem I have with my i7 930 - it always averages out around 40'C idle no matter what I do in a 70'F room...even though other people report idles of 30-35'C with the stock cooler and higher room temps. Again, the temps are in the lower 30s when resuming from sleep mode.
m
0
l
a b à CPUs
October 3, 2010 6:52:19 AM

Temps can be affected by many things. Flatness (even at the microscopic scale) of the heatsink base and CPU heatspreader, type and amount of TIM used, heatsink used, mount pressure, room temperature, case temperature and airflow, etc. At this end of the temperature scale you also have to take into account increased amounts of sensor inaccuracy. I wrote something on it here** about a year ago. Don't pay too much attention to the stuff just prior to section 3. I think there's an issue with it that I didn't get around to correcting, and I'm too rusty on this subject to fix it.

**I must have been half asleep while writing some of this. My grammerz in the first paragraph is quite poor.
m
0
l
a b à CPUs
October 3, 2010 6:57:29 AM

Since I have all other factors (HSF, TIM, room temps, case, etc) at ceteris paribus, I guess you're right that my problem probably stems from inaccurate temperature sensors.

Damn you poor quality 45nm sensors!!! >_<
m
0
l
a b à CPUs
October 3, 2010 7:14:14 AM

As long as they trip thermal protection at the right time they are doing their job.
m
0
l
a b à CPUs
October 3, 2010 7:30:09 AM

The other thing is the operating system and processes running in the background.

If you swapped the 6600 for the 9450 and the system config is exactly the same then fine (i.e., you measured temps then swapped and measured temps again).... if your referring to temps taken some time previously then you may have had some processes running in the background which would have lifted the cpu slightly above idle for the 9450 ... accounting for the temp variance.

The 45nm temp sensors are inferior to the 65nm ones - fact.

Use a pizza oven laser if your not sure ... seriously the 45nm cpu is probably chewing less current under load / as well as at full load.

I have a Q6600 here and an E5300 ... I swap them around a couple of systems depending on what I need at the time ... I ignore the E5300 temps ... though coreTemp seems pretty good as a utility.

The 9450 is about 9% faster ... on IPC across a range of benchies ... so its a better proprosition to run.

http://ark.intel.com/Product.aspx?id=33923

Push the FSB up to the point where it is stable without having to lift the CPU core voltage ... find the sweet spot so to speak.

Since stock it runs 333FSB (quad pumped to 1333) X 8 = 2.66 then 400 (1600) should give you 3.2Ghz which is easily achieved I'd imagine at stock voltage.

Good luck.

m
0
l
October 3, 2010 8:10:58 PM

Stuck sensors as was stated more precisely above. I have the same problem on my Yorkfield. Trust me, save yourself the hassle of what i did which was to reseat my heatsink/fan 5 times.

Basically, go by your lowest idle core temp if you're worried. My sensors are stuck at 39, 28, 38, 41. Therefore I know that 28 is more likely where the temp really is since that is the only one to ever fluctuate below its current number.
m
0
l
a b à CPUs
October 3, 2010 8:46:12 PM

Check out the thousand page thred on real temp, it's pretty informative. The author of realtemp is quite interactive in the thred, answering most questions. He and others explain the 45nm core2 senser issues.
m
0
l
!