Temp Problem with Q9450

[Jonilink]

I have a few issues and I hope I'm in the right sub forum I recently put together a new system:

EVGA 780I P04
Q9450 D-tek fuzion cpu block (stock nozzle)
Microres 1/2 tygon MCP-655b pump

got everything running bumped the FSB up to 400 and got and easy 3.2ghz
with out touching any other setting in the bios. now here's the problem

The temp readings vary ALOT from core to core (see pic)

although both programs show different temps you can see that cores 1 and 2 are over 10c-15c higher than cores 0 and 3.

On my D-tek fuzion I have it tightened down till the springs are almost completly compressed and applied a thin layer of as5 on both sides.

Does the quad nozzle fix this problem or should I just reapply TIM and reseat my block? I have the nozzle kit, but dont really want to tear it all down unless I really have to.

 

[Shadow703793]

Could be many reasons but the common ones are:

1. CPU Load for each core is different
2. The CPU heat spreader is not very smooth/flat (ie. too concave/convex)
3. Bad DTS sensors

 

[Jonilink]

ok the cpu load was the same for all cpus: none this is idle I waited a bit after the os loaded to make sure nothing was running

so if the cpu heatspreader is to concave/convex lapping is the only solution?

bad DTs sensors I was just reading up on this and I ran the Realtemp test and it's giving me all sorts of numbers for example

im getting
11
6
5
10

then
9
5
6
9

I've reseated the block twice today and after the second time I put a thin cost of AS5 on the cpu and block and it brought the temps down a little.

 

[elbert]

I suggest you check your heat sink face for flatness. If its concaved you may need to lap. Also check your CPU. Make sure when reapplying the tim its as thin as possible while completely covering the CPU face.

 

[Jonilink]

im considering an RMA of this cpu just checked two other quads and they do not have this problem being so far/high on temps with my wc setup.

I reseated my cpu and this time used the zalman brush to apply a thin coat to just the cpu.... no difference

I was also wondering if the orientation of the plate makes a difference (see pic)

 

[CompuTronix]

jonilink, you've already identified the problem. Your results from Real Temp "Test Sensors" suggests that Cores 1 & 2 have DTS sensors which are "sticking" at Idle. This is a common problem with 45 nanometer processors. From the Core 2 Quad and Duo Temperature Guide - http://www.tomshardware.com/forum/221745-29-core-quad-temperature-guide


Section 15: Troubleshooting

Notice: A significant percentage of 45 nanometer processors are being reported with defective DTS sensors, which appear as unresponsive Core temperatures at lower Scale. Offsets between Cores exceeding 10c are also being reported. Sensors can be tested using Real Temp 2.5 - http://www.techpowerup.com/realtemp/ - Processors with defective sensors should be RMA`d. This applies to E7000, E8000, Q9000 and QX9000 series processors.


Hope this helps,

Comp

 

[Jonilink]

yah got the rma all setup good thing I didnt lap the cpu thx for all the help !

 

[CompuTronix]

You're very welcome.