Q6600 Not-yet-OCed Temps

[MoeStooge]

I was planning on OCing my Q6600 system this week (put it together before X-mas and haven't had the chance to mess around yet).

I downloaded SpeedFan and noticed that the temps displayed for each of the cores at idle are 19, 17, 11, and 11. The 6-8 degree difference between cores 2/3 and 0/1 seemed a little odd to me and has me wondering if I did not get the thermal paste applied evenly across the cores. Is my reasoning good or is this something that normally happens?

I am using the TT Ultra Extreme and a Scythe 120mm fan.

 

[VirtualLarry]

That large a core varience across cores is normal for a quad-core.

 

[sharad]

Have you tried using Core Temp to monitor the temps?

 

[BonzaiDuck]

Either you are freezing in your room at temperatures between 53 and 66F, or SpeedFan may not be giving you reliable readings.

Use CoreTemp. Then, show us again the idle temperatures for the 4 cores. And let us see your room-ambient temperature . . .

 

[Idontcare]

Originally posted by: BonzaiDuck
Either you are freezing in your room at temperatures between 53 and 66F, or SpeedFan may not be giving you reliable readings.

Use CoreTemp. Then, show us again the idle temperatures for the 4 cores. And let us see your room-ambient temperature . . .

This is a rather well known, well documented, and easily corrected configuration issue with Speedfan and G0's (it is off by exactly 15C).

See section 5 in the middle of Graysky's OC guide (complete with instructions on how to correctly configure Speedfan): http://forums.anandtech.com/me...=2057083&enterthread=y

@OP: I would not settle for such a large die-to-die temp variation if I were you.

I just setup 5 G0 Q6600 systems this past week and every time one of them had a temp delta >4C I removed the HSF, cleand, and re-applied AS5 and sure enough I could make the temperature delta drop to <3C for all the systems.

The temperature differential is definitely a function of how well the HSF is coupled to the IHS. On two of my systems I went from a 4C temperature delta to 0C delta by reseating my HSF.

 

[VirtualLarry]

Originally posted by: Idontcare
The temperature differential is definitely a function of how well the HSF is coupled to the IHS. On two of my systems I went from a 4C temperature delta to 0C delta by reseating my HSF.

Or how well the IHS is shaped. If it is uneven, then lapping may help, but re-seating the heatsink likely wont.
Requiring a 0C delta is ideal but highly unrealistic. Most people have 6-7C deltas between cores on a QC.

 

[BonzaiDuck]

It seems to vary according to % CPU-Usage between idle and load. No -- I'll elaborate. It varies by % CPU-Usage and by BIOS "set" voltage. And the "monitored" voltages would vary the same way, as there just seems to be a constant bias between "set" and "monitored."

So at an over-clock of 3.0 Ghz, 1.31875V VCore, the idle range among four cores is 3C degrees and the load range among cores is between 5 and 6C degrees.

With an over-clock of 3.15 1.4183V the idle range among four is more like 4 to 5C, and the load range can reach 7C degrees.

And I'm using diamond thermal paste with a 95% loading.

Keep in mind I'm providing these measurements using a B3 stepping of Q6600, but I suspect that at 3 Ghz, both steppings require about the same voltage. At other speeds, it probably differs between the two.