High uneven temps on liquid cooled Core 2 Extreme QX9650 (cores 0 & 3)

Hello. New to Tom's (or forums in general) so I hope this posts correctly.

I have a QX9650 that is giving me temp problems. The CPU is on liquid and by itself in the lupe @ stock clock and voltage. Idle temps are:


Core 0=50c Core 1=30c Core 2=38c Core 3=52c


These temps are reported about the same in every monitoring prog I use under both XP and Vista. I'm currently using core temp 0.98.1. I thought the block/mounting may be at fault so I installed a Q6600. The temps for the Q6600 were idle at about 35c each core. I then lapped my QX9650 assuming it was the CPU's spreader causing the uneven temps. That did little to nothing.

My question is: Could this be an internal issue with the cores not hitting the spreader? Or is it something to do with my setup?

Case: Lian Li V1200 b Plus II
PSU: Silverstone 1200W
MOBO: ASUS Striker II Formula 780i
CPU: QX9650 on Swiftech Apogee GT (Bowed), MCR120 Rad, MCP355 pump
GPU: 2x 8800GTX on Swiftech Stealth GTX, MCR220 rad, MCP 355 pump
RAM: 2x 2GB OCZ HPC DDR2 1000
HDD: 2x 150GB Raptors + Hitachi 1TB in 4 bay backplane
Optical: Philips BDR Drive

The warranty is gone due to lapping so I hope someone has an idea or solution. Any help is appreciated. Thanks!

Hatman, have you tried your current cooling apparatus on another CPU? If so, does it report better temps?

Stock heatsink was worthless. I am using real temp 2.6 under XP and getting 41c 41c 28c 20c. The cores are not assigned. Are they in order 0-3? In any event, the temp sensors seem to be stuck. I got as stated below:

Core 1: Sensor Movement: 4
Core 2: Sensor Movement: 4
Core 3: Sensor Movement: 0
Core 4: Sensor Movement: 10

This shows that core 3 is stuck and that core 4 is inaccurate (if not stuck) according to the real temp explaination. So anyone who is having the same issue, this is why. This is specifically effecting your idle temp readout.

Thanks pcgamer12.... This solved the mystery. Any idea on how to get them unstuck?

stuck sensor only RMA can be the solution.

I don't think they will RMA a lapped CPU.

LOL... I'll work on the nickel plating! I wonder if they'll be able to tell? I think I'm satisfied though. At least I know the temps aren't crazy high. Time to start overclocking the hell out of it now!

Yep,

Stuck sensors.

My first Q9450 did nearly the same thing, RMA'd her for another, but after lapping, you cannot do that.

$1000 paper weight now.

Real Temp is the only thing I know that reads the temps correctly.

thats why im still not so confident lapping a CPU without running it and OC it to the limit and confirm its "SAFE". good luck on your way to OC lan!

I ran p95 for a few hours and the temp sensor is now responding... Even when idle!

I've read that only the idle temps are affected by the stuck sensors so I ran p95 and was getting pretty even temps. I let go for about 10 hrs and, when I stopped, the sensor was responsive even when idle.

Got the cpu oc'd to 4GHz (400x10) @ 1.375v. Idle temps are:

Core 1: 35c
Core 2: 33c
Core 3: 32c
Core 4: 36c

Each core hits the mid 60's under full load w/ p95 small FTTs. For a 4GHz quad, I'm pretty satisfied with those numbers!

I put a post in the overclocking section about cpu intelligent accelerator found on the x48 gigabyte boards. I think its fantastic, you dont have to disable anything and it steps down when your not using power and still oc's really well. Mine will jump to 3.9ghz if needed, render anything and run crysis without ever dropping out. I think people should look at optimising a system rather than just crank the fsb and cpu output. I hope dynamic overclocking become automatic one day, you either enable or disable it and get max power when you need it, and save energy when you dont.