Are-My-CPU-Temperatures-High???

[Zer0Cool]

Hello,

The other day I figured out how to properly use Prime95 to stress all 4 cores, (My original thread had false temps) here are my results using torture test on all 4 cores using small fft's for 10min on my Q9450 (stock speed 333 x 8 = 2.6Ghz) all my specs are in my sig.

Ambient = 22c

SpeedFan v4.34

Idle
CPU = 18c
Core 0 = 32c
Core 1 = 31c
Core 2 = 44c
Core 3 = 43c
CPU-Z = 1.232V
CPU-Z Vcore (Idle) = 1.232V

Load
CPU = 33
Core 0 = 47c
Core 1 = 46c
Core 2 = 57c
Core 3 = 55c
CPU-Z Vcore (Load) = 1.200V

I have an “Ultima 90” as my heatsink with 92mm Noctua fan inside a Cosmos S case with 6 120mm fans. Heatsink is lapped as well as IHS of Q9450. Looking at the temps you can see that there is about 12c difference between cores 0,1 & 2,3. I tried reseating the heatsink & I tried using stock Intel cooler, but the difference in temps stayed the same. (RealTemp gives the same 12c difference)

What cores do u think are giving proper temperature? Let’s assume its cores 0,1 is 47c on full load at stock is high? How far do u think I will be able to OC this processor? I am very upset about all this, obviously I can’t return the CPU now that I lapped it, and spending $360 on a new one doesn’t sound 2 appealing. What are your general thoughts and suggestions?

Thanks.

 

[harbin]

Temp is ok to me.

 

[wtburnette]

No. See my response and other's here:
http://www.hardforum.com/showthread.php?t=1304232

 

[aethelwulf]

It looks alright, but with that much difference between cores, you may need to reseat your HSF.

 

[Nenu]

Looks to me like some of the sensors are stuck, its a known issue.
You wont get that much difference between cores from badly applied heatsink paste, all core temps would rise.

There is a possibility that the paste on the underside of the core cap isnt covering the whole chip but as the temps arent that high anyway, I doubt it.

 

[a104375]

yeah those temps all look normal, nothing out the the norm, heck ive seen my computer with temps higher than that

 

[vanilla_guerilla]

give it some volts and overclock it. unfortunately, q9450 are not known as decent overclockers, and that 8x multi is going to kill you unless you give cpu and nb some juice. also, unless you have an nvidia chipset, you will need fast ram. should have got the q6700 for $270. that 12mb l2 isnt going to mean much at 3ghz, when you compare it to a q6700 at 3.8 to 4.0ghz. whatever made you decide to buy one of those? immense crowds of people are avoiding that chip. on the day it was released huge throngs of enthusiasts did not buy that cpu. if you check the fs/ft forums you will see that hundreds of overclockers are not selling that proceesor, mainly because they didnt buy one. but seriously, if you check out the xtremesys intel forums there are dozens of threads that have nothing to do with that cpu.

 

[Zer0Cool]

It looks alright, but with that much difference between cores, you may need to reseat your HSF.

Thanks this is what RealTemp is giving me;

Idle
Core 0 = 25
Core 1 = 25
Core 2 = 39
Core 3 = 37

Load
Core 0 = 40
Core 1 = 40
Core 2 = 51
Core 3 = 50

Looks to me like some of the sensors are stuck, its a known issue.
You wont get that much difference between cores from badly applied heatsink paste, all core temps would rise.

There is a possibility that the paste on the underside of the core cap isnt covering the whole chip but as the temps arent that high anyway, I doubt it.

I used the credit card method this time so the paste should cover the entier chip

yeah those temps all look normal, nothing out the the norm, heck ive seen my computer with temps higher than that

hmm I was almost sure my temps are high! i really want to belive that cores 3-4 are stock and not giving propper numbers

give it some volts and overclock it. unfortunately, q9450 are not known as decent overclockers, and that 8x multi is going to kill you unless you give cpu and nb some juice. also, unless you have an nvidia chipset, you will need fast ram. should have got the q6700 for $270. that 12mb l2 isnt going to mean much at 3ghz, when you compare it to a q6700 at 3.8 to 4.0ghz. whatever made you decide to buy one of those? immense crowds of people are avoiding that chip. on the day it was released huge throngs of enthusiasts did not buy that cpu. if you check the fs/ft forums you will see that hundreds of overclockers are not selling that proceesor, mainly because they didnt buy one. but seriously, if you check out the xtremesys intel forums there are dozens of threads that have nothing to do with that cpu.

Well I wasnt really planing on going for 4ghz here....but thank you for letting me know what a bad decision I made, I feel much better now

 

[Nenu]

I used the credit card method this time so the paste should cover the entier chip

That wont make a difference, I was referring to the paste between the CPU and the CPU cap that is put there at the time of manufacture.
As pointed out thats unlikely to be the issue as well.
(The cap is the metal cover that the heatsink contacts with)

 

[Zer0Cool]

That wont make a difference, I was referring to the paste between the CPU and the CPU cap that is put there at the time of manufacture.
As pointed out thats unlikely to be the issue as well.
(The cap is the metal cover that the heatsink contacts with)

ic what u mean now, ya who knows...

I dont know should I bother with reseating the HS? Like I sad b4 I tryed doing it b4, didnt work. I also tryed the Intel HS, the difference was the same, so its not the HS. If like u guys say the RealTemp is giving correct info and Cores 3,4 are stock so that would mean I have 40 under load, which is good right? What u think?

 

[vanilla_guerilla]

Well I wasnt really planing on going for 4ghz here....but thank you for letting me know what a bad decision I made, I feel much better now

absolutely no problem. if you check in here whenever you have made a choice i will be happy to let you know, again.

 

[delvryboy]

Depending on your clock, expect mid 50's under P95.

You may have a stuck sensor, but they usually work under load. In most cases, large differences between the cores is a sign of a convex IHS.... lapping may be in order.

 

[RamonGTP]

Looks to me like some of the sensors are stuck, its a known issue.
You wont get that much difference between cores from badly applied heatsink paste, all core temps would rise.

There is a possibility that the paste on the underside of the core cap isnt covering the whole chip but as the temps arent that high anyway, I doubt it.

How are they stuck? Every core changed by well over 10 degrees between idle and load.

 

[Nenu]

Soz, I missed earlier that they changed.
Its also a common issue for some cores to register temps that dont match other cores.
Pot luck what you get I'm afraid.

I have come across talk of unsticking some of the stuck sensors.
On a similar note, maybe heat cycling or extremes of termperature for a period of time can affect bad calibration positively?
Best to read what others have tried though to be sure not cause damage.

 

[Astral Abyss]

I'm a bit baffled that you would consider a RealTemp load temp of 51C on your hottest core to be "bad".

My Q6600 G0 has core temps that are hotter by 8-9C on 2 of the cores. I lapped the CPU (which was very concave... lower in the middle) and it made no difference in the temperature variance. I'm using the same heatsink as you, but I was using a Tuniq Tower previously and the results were similar. You have to figure that since the quads are bascially just 2 duos slapped together that you're going to get differences like that on a lot of them. Also, we have no idea how well the heatspreader is making contact with the cores on the backside and no way to fix it besides removing it.

As long as you keep your temps under 70C you should be fine. Even that's not really excessive for a quad. People get all paranoid about temps, yet they forget that the crappy stock heatsink doesn't cool very well and that Intel expects the temps to get pretty high, which is why Intel can get away with using such a junk heatsink.