Quad core overclocking questions-Q6600

[Ausm]

I am just getting back into the overclocking scene after a long hiatus and I have some questions.

Here is what I got my rig up to.

Q6600 g0 9 X 367 3.3ghz with a vcore of 1.3815

Idle temp c1 30C C2 29C C3 26C C4 30C

Full load temp C1 51C C2 50C C3 47C C4 51C

Ran prime 95 for 18 hours with no errors

Here are my questions:

1) Is it normal for 1 or more cores to run lower temps on these CPU's?

2) Is it normal to have your BIOS Vcore set for 1.4 but Speedfan reads the actual Vcore as 1.3815

3) Opinion on my overall temps?

TIA,

Ausm

 

[somethingsketchy]

My two cents on your questions:

1. It is not uncommon to have one or two cores be lower than the other two or three cores. My Q6600 (running Rosetta@home 24/7) has Core #2 posting a 4 degree Celsius difference from Core #1 and Core #3. I wouldn't be too shock at that little of a difference, unless your temperature difference is like 10+C from one core to another.

2. I am not sure about SpeedFan (never used it) but some software have a habit of under- or over-reporting voltages. Hell some motherboards may even under- or over-volt the CPU (learned that lesson from a friend), I would go with what your BIOS tells ya.

3. Those temps are looking pretty damn good. Granted if you intended to run for 24/7 at 100% CPU usage I would be cautious but not worried. Intel's Thermal Specification...
http://processorfinder.intel.c...tails.aspx?sSpec=SLACR
...states that the q6600 has a "limit" of 71 degrees Celsius. Now personally I never let my CPU get near 55C, but for longevity reasons.

 

[BonzaiDuck]

Originally posted by: somethingsketchy
My two cents on your questions:

1. It is not uncommon to have one or two cores be lower than the other two or three cores. My Q6600 (running Rosetta@home 24/7) has Core #2 posting a 4 degree Celsius difference from Core #1 and Core #3. I wouldn't be too shock at that little of a difference, unless your temperature difference is like 10+C from one core to another.

2. I am not sure about SpeedFan (never used it) but some software have a habit of under- or over-reporting voltages. Hell some motherboards may even under- or over-volt the CPU (learned that lesson from a friend), I would go with what your BIOS tells ya.

3. Those temps are looking pretty damn good. Granted if you intended to run for 24/7 at 100% CPU usage I would be cautious but not worried. Intel's Thermal Specification...
http://processorfinder.intel.c...tails.aspx?sSpec=SLACR
...states that the q6600 has a "limit" of 71 degrees Celsius. Now personally I never let my CPU get near 55C, but for longevity reasons.

I concur with all this, but I've undergone a revision in my temperature concerns -- which used to reflect those of Sketchy there. You'll do more damage with voltage increases than with higher thermals -- with caveat that higher voltages lead to higher thermals. In both cases, it's a good idea to keep the extremes within spec.

Unless the temperature disparity between the core readings is due to a variation in the accuracy of the sensors, you might be able to reduce both the overall temperatures and the difference between them by lapping off the nickel-plate on the processor cap (IHS) -- which also makes the surface flatter. if there is nickel-plate on the base of your heatsink, lap that off too. I'd guess that this -- plus the use of diamond TIM to replace something like Arctic Silver 5 -- could drop the load temperatures 5C to 10C degrees.

 

[F1shF4t]

Originally posted by: somethingsketchy
My two cents on your questions:

1. It is not uncommon to have one or two cores be lower than the other two or three cores. My Q6600 (running Rosetta@home 24/7) has Core #2 posting a 4 degree Celsius difference from Core #1 and Core #3. I wouldn't be too shock at that little of a difference, unless your temperature difference is like 10+C from one core to another.

2. I am not sure about SpeedFan (never used it) but some software have a habit of under- or over-reporting voltages. Hell some motherboards may even under- or over-volt the CPU (learned that lesson from a friend), I would go with what your BIOS tells ya.

3. Those temps are looking pretty damn good. Granted if you intended to run for 24/7 at 100% CPU usage I would be cautious but not worried. Intel's Thermal Specification...
http://processorfinder.intel.c...tails.aspx?sSpec=SLACR
...states that the q6600 has a "limit" of 71 degrees Celsius. Now personally I never let my CPU get near 55C, but for longevity reasons.

Agree with most of these points. Just a few things i'd add.

1. Its normal to have a slight difference in core temps. One of my q6600 has one core which is always 3-5 C lower than the rest, this evens out to same temps at full load.

2. QFT. Also there will be always some sort of voltage droop. Real temp or Coretemp are much better progs for reporting cpu temps, op should get one of em.

3. I always like it when everyone reports these great temps, only problem is that they are prime 95 temps or game temps etc. Run some Linpack, Linx or intel burn test for a couple of hours and see those temps go A LOT higher.

The 71C temps are TCase specs, core temps are determined by a delta from TJMax. In the case of Q6600 TJmax is 100C. Keep this delta higher than 20C (or core temps below 80C) under linpack and you should be fine. Why linpack?, cause then you know that even 24/7 load of something like folding etc will not ever get close to those temps. (100% Prime 95 temps should be 10-15C lower than linpack temps)

BTW: The delta between core temps and Tcase temp is usually 5-10C. Throttling temps for these chips is at 95C or 5C from TJMax and thermal shuttoff should happen when TJMax temp is reached.

 

[BonzaiDuck]

Originally posted by: Dark Cupcake

Originally posted by: somethingsketchy
My two cents on your questions:

1. It is not uncommon to have one or two cores be lower than the other two or three cores. My Q6600 (running Rosetta@home 24/7) has Core #2 posting a 4 degree Celsius difference from Core #1 and Core #3. I wouldn't be too shock at that little of a difference, unless your temperature difference is like 10+C from one core to another.

2. I am not sure about SpeedFan (never used it) but some software have a habit of under- or over-reporting voltages. Hell some motherboards may even under- or over-volt the CPU (learned that lesson from a friend), I would go with what your BIOS tells ya.

3. Those temps are looking pretty damn good. Granted if you intended to run for 24/7 at 100% CPU usage I would be cautious but not worried. Intel's Thermal Specification...
http://processorfinder.intel.c...tails.aspx?sSpec=SLACR
...states that the q6600 has a "limit" of 71 degrees Celsius. Now personally I never let my CPU get near 55C, but for longevity reasons.

Agree with most of these points. Just a few things i'd add.

1. Its normal to have a slight difference in core temps. One of my q6600 has one core which is always 3-5 C lower than the rest, this evens out to same temps at full load.

2. QFT. Also there will be always some sort of voltage droop. Real temp or Coretemp are much better progs for reporting cpu temps, op should get one of em.

3. I always like it when everyone reports these great temps, only problem is that they are prime 95 temps or game temps etc. Run some Linpack, Linx or intel burn test for a couple of hours and see those temps go A LOT higher.

The 71C temps are TCase specs, core temps are determined by a delta from TJMax. In the case of Q6600 TJmax is 100C. Keep this delta higher than 20C (or core temps below 80C) under linpack and you should be fine. Why linpack?, cause then you know that even 24/7 load of something like folding etc will not ever get close to those temps. (100% Prime 95 temps should be 10-15C lower than linpack temps)

BTW: The delta between core temps and Tcase temp is usually 5-10C. Throttling temps for these chips is at 95C or 5C from TJMax and thermal shuttoff should happen when TJMax temp is reached.

Alfredo Comparetti has shown himself to be a darn good programmer, and SpeedFan at one time was truly great for thermally controlling my CPU fan. But I don't trust the temperature readings from the program anymore -- so -- yeah -- CoreTemp or RealTemp -- the only way to go for that . . .

BTW: The delta between core temps and Tcase temp is usually 5-10C. Throttling temps for these chips is at 95C or 5C from TJMax and thermal shuttoff should happen when TJMax temp is reached.

Probably no need for me to add this caveat: that the tCase-to-core spread you mention seems to be specific to the Conroe/Kentsfield cores. There are all sorts of anomalies with the Wolfdales and Yorkfields. It almost seems to me that my Wolfdale TCase is about equal to the core values -- when latter are high enough under load so that I can read them from the "stuck" sensors.

 

[VirtualLarry]

Originally posted by: Dark Cupcake
BTW: The delta between core temps and Tcase temp is usually 5-10C. Throttling temps for these chips is at 95C or 5C from TJMax and thermal shuttoff should happen when TJMax temp is reached.

Not quite true. throttling occurs when TJ = 0, means temps hit TJmax. Thermal Shutdown does not occur until the BIOS temp shutdown temp triggers.

(Yes, I've tested this recently, that's how I know.)