Need help stablizing Q6600 B3 at 3.2Ghz

superclocked

Member
Aug 19, 2007
36
0
0
Hey

I am new to overclocking. This the first time I built my own computer and I am trying to stabilize my Q6600 at 3.2Ghz.

My idle temp is around 37c and on full load it's about 57c. The only problem is when I run prime95 (Quad-core Stress edition) I get two different full load temperatures. When It processing big FFT I get max load of 57c with all cores around 68 but when it start processing small FFT my cpu temperature raise to 70c and all cores around 79c.

b.t.w - I am using Everest to measure these temperatures.

My Question is are these temperatures normal?

Q6600 B3 - 3.2Ghz - 1600x8
Cpu cooler - Air = Freezer pro 7
motherboard 680i
8 Gig of ram timing 5-5-5-12 400Mhz

Voltages that I changed
CPU Voltage = 1.4 Bios --- cpuz Voltage =1.36
FSB voltage = 1.4 Bios

Thank you.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
27,392
16,236
136
Yes, you need better cooling. for 3.2 on a quad, many recommend the Ultra 120 cooler.
 

superclocked

Member
Aug 19, 2007
36
0
0
Thank you for replying.

Will ultra 120 fit in lexa case (mid tower)? I have 92 mm fan now and 120mm fan behind the heat sink.
 

superclocked

Member
Aug 19, 2007
36
0
0
I have an nvidia 7900 GT KO 512 and antec True power 850watt power supply.

new temps I have on

Prime95 small fft:

cpu= 69
Core1 = 80
Core2 = 79
Core3 & 4 = 78

Prime95 big fft:

cpu = 56
core1 = 70
core2 = 69
core3&4 = 68

I have been running prime95 about 4 hours now.
 

graysky

Senior member
Mar 8, 2007
796
1
81
Your temps are too high dude. Like others have said, you'll need better cooling and good case ventilation. Also consider lapping the IHS and base of the HS. Also, your vcore is near the redline for the chip (1.372 V) but you should know that more mV you pour into it, the greater the heat production. It's actually an exponential increase.

Read this thread for more on this.
 

superclocked

Member
Aug 19, 2007
36
0
0
I just ordered the ultra 120 today. I hope It fits in the mid tower case.

according to speedfan my temp when running small FFT:
cpu = 71
core1 = 66
core2 = 66
core 3&4 = 63

according to speedfan my temp when running big FFT:

cpu = 61
core1 = 58
core2 = 57
core3 & 4 = 56
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
70C is the throttle point on Q6600s anyway.

Youre losing performance when you cross the 70C mark because the cpu clocks itself down to a x6 multiplier.
 

superclocked

Member
Aug 19, 2007
36
0
0
all the speed limitation features are turned off even after 70c cpuz shows the multiplier at 1600X8 3.2Ghz.

What are max temp Q6600 can handle for the CPU and Cores?
 

jawknee530

Senior member
Jan 16, 2007
213
0
76
i believe that Intel rates the B3 have a 62c limit. i wouldn't want to run my cpu over Intel's rating for very long.
 

stevty2889

Diamond Member
Dec 13, 2003
7,036
8
81
Originally posted by: Acanthus
70C is the throttle point on Q6600s anyway.

Youre losing performance when you cross the 70C mark because the cpu clocks itself down to a x6 multiplier.

From my own testing, throttling doesn't kick in until it reaches the Tjunction temp. When I was using a Scythe Ninja, I was up to 85c and not throttling..the Tjunction is 100c.

The Tjunction for an E6600 is 85c. My roomate got his running that hot, and thats where it started throttling.

Of course I don't think it's a good idea to let it run that hot either, which is why I switched to water, but even with water I can't get my B3 stable at 3.2ghz. Then again I couldn't get my E6600 stable at 3.2ghz on the same motherboard.
 

stevty2889

Diamond Member
Dec 13, 2003
7,036
8
81
Originally posted by: superclocked
I just ordered the ultra 120 today. I hope It fits in the mid tower case.

according to speedfan my temp when running small FFT:
cpu = 71
core1 = 66
core2 = 66
core 3&4 = 63

according to speedfan my temp when running big FFT:

cpu = 61
core1 = 58
core2 = 57
core3 & 4 = 56

You have to add 15c to those core temps in speedfan to be accurate. It calculates the Tjunction as 85c, when it's actualy 100c. So you are really running at 71-73c.
 

graysky

Senior member
Mar 8, 2007
796
1
81
stevty is right... use coretemp which doesn't need any correction. Also, lap that chip and the base of your HS.
 

superclocked

Member
Aug 19, 2007
36
0
0
Thank you guys all this good to know before I fry my Q6600.

The only reason I increased my vcore to 1.4v was at 1.350 and 1.375 one of core failed about 2 hours and 45 min(s) so I increased voltage.

one more thing what are acceptable temps and voltages for cpu and cores when overclocking to 3.2Ghz when measured with coretemp or everest?


just a thought:

sometimes overclocking act really strange just yesterday I ran the everest system stability test for 8 hours at 3.2Ghz with Vcore at 1.334v max temp according everest for cpu was 60c with cores around 68c but prime95 failed after running 5 hours.
 

SickBeast

Lifer
Jul 21, 2000
14,377
19
81
You could try lapping.

With those temps I would run it at 3.0ghz like most people with B3's are doing.
 

superclocked

Member
Aug 19, 2007
36
0
0
ok I am getting the ultra 120 heat sink. I heard its the best if you are on air cooling.
About lapping the current heat sink I could only find up to 600 Grit paper in the local store. It's labeled ultra fine and meanwhile I lowered the vcore. now its at 1.386 I am running the small fft on prime according to everest.

small fft 8K

CPU temp = 66
core 1= 78
core 2= 77
core3 & 4 =73

a little better than before.
 

ChronoReverse

Platinum Member
Mar 4, 2004
2,562
31
91
Just to be clear, the TJunction temperature, by definition, is the point where the chips start to throttle.
 

Aznguy1872

Senior member
Aug 17, 2005
790
0
0
If all else fails and you can't get it stable at 3.2 Ghz, just bring it down to 3.0 Ghz and lower the voltage.
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: ChronoReverse
Just to be clear, the TJunction temperature, by definition, is the point where the chips start to throttle.

This is correct, i was wrong that the Tjunction is 70C.

Im not sure where i got that number from, maybe the old P4C and P4E days. :eek:
 

PCTC2

Diamond Member
Feb 18, 2007
3,892
33
91
what ambients are you running in?

definitely lap the Ultra120 and Q6600 IHS. I have an Ultra120 on my Q6600 B3 @ stock and I haven't gotten around to lapping it and I have pretty mediocre temperatures (I have reseated it and reapplied paste at least 4 times to test that).

In 31C Ambients:
Idle:
CPU: 42
Cores: 39-44

Orthos Small FFT Load:
CPU: 60
Cores: 59-63
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: PCTC2
what ambients are you running in?

definitely lap the Ultra120 and Q6600 IHS. I have an Ultra120 on my Q6600 B3 @ stock and I haven't gotten around to lapping it and I have pretty mediocre temperatures (I have reseated it and reapplied paste at least 4 times to test that).

In 31C Ambients:
Idle:
CPU: 42
Cores: 39-44

Orthos Small FFT Load:
CPU: 60
Cores: 59-63

Youre right in the normal range.

You are essentially cooling 2 dual core CPUs with one heatsink remember :p
 

BonzaiDuck

Lifer
Jun 30, 2004
16,864
2,183
126
The original Ultra-120 is fully adequate, but the U-120-Extreme should give you as much as a 5C-degree advantage. And if you're OC'ing, you shouldn't worry about the price differential.

It seems even I have become confused about the temperatures. I think I'm right, though, that "TCase" is the reference point for the thermal throttling limit, but this would be consistent with the TJunction value as well. Thus, when TCase reaches 63C for the E6600, the processor will throttle. For the E6700 and others, it would be 65C. I'm not absolutely sure what it is for the C2Q (and darn -- I should know) but I think it's also in that ball-park.

The core temperatures should show load values about 15C higher than TCase with the C2D processors. I've observed a spread of only half that for the C2Q.

The TCase temperature is the "overall" CPU temperature -- for a 680i board, it is the CPU temp reported by nVidia Monitor.

Everest Ultimate gives accurate readings on the cores, but it's got a lot of overhead due to its extensive monitoring activity, and biases the Core 0 temperature upward a few degrees. Correct me if I think this is a little like the Heisenberg Uncertainty Principle . . .

Forget SpeedFan for a 680i system, and use CoreTemp. If you're running Vista, EU may be your only other option.
 

superclocked

Member
Aug 19, 2007
36
0
0
Originally posted by: PCTC2
what ambients are you running in?

definitely lap the Ultra120 and Q6600 IHS. I have an Ultra120 on my Q6600 B3 @ stock and I haven't gotten around to lapping it and I have pretty mediocre temperatures (I have reseated it and reapplied paste at least 4 times to test that).

In 31C Ambients:
Idle:
CPU: 42
Cores: 39-44

Orthos Small FFT Load:
CPU: 60
Cores: 59-63


well today I am 22c ambient about 72F. I am running prime95 now with lower vcore (1.29) than before. temp are as fellows:

prime big fft load:
Everest
cpu = 51
core 0 = 65
core 1 = 62
core 2 & 3 = 61

prime small fft load:

Everest
cpu = 64
core 0 = 75
core 1 = 75
core 2&3 = 72

my core 0 is always the hottest one but BonzaiDuck said something about everest being biases against core 0.

Got a question If one of the core fails on prime95 that means your system isn't stable and I should increase vcore?

vcore = 1.356v in bio
vcore = 1.29v in windows when underload

memory voltage is set auto at 1.85v could this cause prime95 to fail? Should I increase this to 2V? which is recommend voltage according to patriot memory.

cpu fsb = 1.2v

What is voltage damping? someone mentioned in another thread on this forum.
 

superclocked

Member
Aug 19, 2007
36
0
0
Does anyone know if Thermalright ultra 120 Extreme and evga 680I A1 Mobo are compatible with each other.

I heard the Thermal ultra's backplate problem with evga mobo. I have an 122-CK-NF68-A1 evga 680I.

Thank you