Intel QX9650 Drastic Core Temperature differences!

[JohnZS]

Hi

I am not sure whether this is normal behavior for a Quad Core CPU or if something is wrong, so please pardom my ignorance on this subject as I have just upgraded from a S478 Northwood!

Anyway here is the story, basically I have a QX9650 running on an ASUS P5E3 (non deluxe and no bells or whistles) running (at the moment) @ Stock speed and voltage.

Both Coretemp 0.95.4 and Everest build 4.2.1233 report a huge difference in temperature between cores 1+2 and 3+4

Basically during stresstesting and/or wPrime 1.58 Core temps are as follows (after 20mins of 100% load).

1 = 49C
2 = 50C
3 = 60C
4= 61C

Any ideas as to why there is a huge temperature difference between the cores?

Idle temps are also different

1 = 31C
2 = 30C
3 = 49C
4 = 51C

My motherboard is flashed to the only BIOS which supports the QX9650 (0703) as the shipping BIOS (403) says "Unkown CPU detected" Please update BIOS from ASUS (although system POSTS, it is unstable)

So, is this normal?
If not any suggestsions?

Thanks

John

[TuKo]

My suggestion : have a look at cooling fixation

[dunn2953]

it is either the heatsink seating, the application of thermal paste, or the ihs on the chip.

[Ace-a-Rue]

i think it is a interpolation problem with the DTS output off each core...i have the same problem at idle and maybe a little worse...could be they don't know how to write the bios code to interpret the readings....even the latest beta coretemp 0.95.6 is not reading correctly.

[Hornet331]

same here, cores are like the following at stock load (@IFX-14) in the first pic attached and on the scythe mugen on the seccond pic.

the heatsink where seated correctly (i have remounted them several times).

maybe the TIM under the heatspreader is the cause of the problem?

[Ace-a-Rue]

here is my 9650 at 4400 mhz for idle and then under wPrime load (1024M bench)

you can see the huge spread in temps for idle conditions but during wPrime 1024M bench they are close enough in my book...it has to be an interpolation problem versus a mounting problem...i am using a TRUE with a strong clamp down.

[xgman]

It is normal even on sub zero cooling. Normal, as in they all seem to do it. Not sure why though. It's probabaly worth one quick check to make sure that you have full coverage though, just to be on the safe side, but they all read waccky like this.

[Ace-a-Rue]

It is normal even on sub zero cooling. Normal, as in they all seem to do it. Not sure why though. It's probabaly worth one quick check to make sure that you have full coverage though, just to be on the safe side, but they all read waccky like this.

EXCUSE ME!...are you trying to say that i don't know how to mount an air cooler?!

[zoob]

Same deal here: 48/46/45/30 @ load :\

[Ace-a-Rue]

coretemp 096 is released...still the same problem:

http://www.xtremesystems.org/forums/...07#post2666507

[AliG]

It is normal even on sub zero cooling. Normal, as in they all seem to do it. Not sure why though. It's probabaly worth one quick check to make sure that you have full coverage though, just to be on the safe side, but they all read waccky like this.

I believe the 45nm cpus have some improved power management, that could be it

[AliG]

EXCUSE ME!...are you trying to say that i don't know how to mount an air cooler?!

I'd say break out the guns and inform the terrorists, we'll hit him with everything we got

[Ace-a-Rue]

i'll get my pee-shooter out of the closet.

[Lestat]

nuttin wrong here
(lol just had to show off the nice cool undervolted lappy cpu)

[Praz]

Check the temperatures at stock settings and see if the difference remains.

[Hornet331]

temps reported by core temp are the same as everest now.

[Ace-a-Rue]

Check the temperatures at stock settings and see if the difference remains.

here are default (stock) settings in idle and load conditions (used wPrime 1024M to load cores).

notice core #0 still reads 51c...doesn't even budge under load.

[Lestat]

i think you can see if it reads 51 idle and 51 load then that shows either the probe on die is broke, or core temp is just reading it wrong.

contact intel and rma it if it bothers you too much.

you also have core 2 that is way off also.

see if TAT works on that cpu it has an onboard stress test so you can see what it reads

[Ace-a-Rue]

TAT does not work; at least the version that i had (already discarded).

AND...core #0 will register above 51c at 4400 mhz when benching wPrime 1024M.

it does not really bother me but i am inquisitive as to why it is this way...that is what i am trying to find out.

[over@locker886]

I was same problem my friend, with his Q6600 G0. I lapped cpu and Scythe Ninja too, but problem stayed. The first core overheating...
..example 3600Mhz 1,36v on load : 90+/73/72/70. At stock speed was same situation, the first core was lot hotter than others...I never saw one like this before it..


If you can, send back to rma..

[Ace-a-Rue]

it is normal when i wPrime it; as normal as can be expected...it's just not accurate when at idle.

i guess i could put in for a RMA...might get back a better overclocker!

[Praz]

At stock/idle all cores appear off unless ambient is that low. Appears as if something is definitely broke.

[brewer265]

my shows bad temps too. Cores 0 and 1 never change.

[JohnZS]

Thanks for the feedback everyone
I checked the mounting of the HSF and all looks well (all for corners are level and the HSF is level, not at an angle).
To me it smacks of 2 things, (having read all the posts), either a BIOS issue (as this processor is relatively new and this is the ONLY BIOS which supports this CPU and even that is not the same version as listed on the ASUS site to support the QX9650) ie I have BIOS 0703 which is the latest version, but ASUS site says that 0704 supports the QX9650. I mailed their Techsupport and they said that the new BIOS was currently in testing.
Idle and Everything Stock

The temperature margins do get narrower during wprime or everest stability test, however Cores 3 and 4 are still reading 10 to 15C higher.
I am sure this will be fixed by a BIOS/Microdode update, if not I shall ask for an RMA or Intel's advice (as I don't want a dead CPU at stock volts and stock clocks!!)
I am going to play it safe and try to resist overclocking until I am happy that the temps are more or less making sense (Less than 10C spread between them).
Thanks for the imput.
What motherboards and BIOS revision do you guys who are having the similar problems have?

I have an ASUS P5E3 BIOS revision 0703

John

[Ace-a-Rue]

i tried my current QX9650 on Abit's IP35 Pro and IX38 QuadGT with the same results...supposedly, IP35 Pro's latest beta bios was supposed to accommodate the QX9650...but...it gave the same basic temp readouts.

on my DFI LT X38 which was recently released, it shows the same thing.

i am also leaning towards a need for micro-code adjustment since many of us are experiencing similar conditions.

EDIT:...just found this showing even temps being read which would indicate that the bios micro-code is correct to handle the 9650.

http://www.xtremesystems.org/forums/...&postcount=294