Overclocking Q6600 healthy temperatures?

[Reven]

Hi

I am currently overclocking my Q6600 and was wondering if I was running healthy temperatures or not.

I overclocked it with a 400 FSB to push it to 3.6ghz. I am running Kingston HyperX DDR2 800mhz ram so I have a 1:1 ratio I believe.

Now, I know you're supposed to do a stress-test, so I did a 32m Super Pi without any issues. I am currently doing a Prime95 test on all my cores.

I have ran Prime95 without any errors for 1 hour and 20 minutes so far. However, after running speedfan and core temp, I realized I may be running very hot.

Here are my temps:

Core 0: 90
Core 1: 87
Core 2: 78
Core 3 : 77

All of these are Celsius.

It should be noted that my I was running even higher temperatures for a while, in response I boosted the speed of my air cooler a bit.

Max temps were:

Core 0: 98
Core 1: 98
Core 2: 83
Core 3: 83

Now, here are my questions: A) Are the lower temps OK? What about the max temps? 98 in particular sounds hot.

B) Why the huge difference in between the cores 1 and 0 and 2 and 3?

Finally, are these temps normal for what I'm using?

I have an Antec 900 with these stats on my fans:
1 120mm intake on medium with 56 CFM.
1 intake 120mm on high with 79 CFM.
1 Intake (side panel) 120mm 90 CFM (though it may be a 69 CFM, I've had conflicting reports)

200mm 'Big Boy' on high 127 CFM

Zalman 9700 at about mid-low settings. I am using the included thermal paste.

Its a bit hot here in Istanbul so the outside temperature is 28C.

Thanks

 

[RussianSensation]

Those are very high temps. My Q6600 crashes at temps above 70*C on any core. I am using Tuniq Tower 120. Have you tried resetting your cooler? 9700 is a great cooler and you should be running much lower at 3.6ghz.

What voltage did you set on the CPU in the BIOS?

Assuming you have a Zalman ZM-TG2 Thermal Grease, its performance is lacklustre: http://benchmarkreviews.com/in...2&limit=1&limitstart=9

If you can pick up Arctic Silver 5 or Tuniq TX-2/3 thermal paste for cheap, I'd substitute the Zalman one. Even then the thermal paste alone isn't enough to account for the temperature difference. Perhaps you applied too much of it?

I tend to think your cooler is not connecting evenly with the heatsink spreader of the cpu since your cores 0 and 1 are significantly hotter.

I don't remember the exact layout of the cores, I think it was:

0 2

1 3

 

[Smartazz]

Those temps are really high Reven. I agree with RussianSensation, reapply your heatsink and thermalpaste and pick up some AS5 or Tuniq TX-2/3 thermal paste if possible. I also run the Q6600 under a Tuniq Tower with Arctic Silver 5 and I definitely would not feel safe with temperatures that high. As a rule of thumb, I try to keep my temps below 60C, which I know is a little conservative, but I just like to stay on the safe side.

 

[Tempered81]

Originally posted by: Reven


Core 0: 90
Core 1: 87
Core 2: 78
Core 3 : 77

All of these are Celsius.


Core 0: 98
Core 1: 98
Core 2: 83
Core 3: 83

Thats a little high. Even for G0. Hopefully it isnt B3

Turn up the fans, decrease the Vcore. Make sure you have a properly seated heatsink, and good case airflow.

 

[error8]

What is the voltage you are pumping into that chip? Those temps are close to throttle, very high and dangerous.

 

[Reven]

Hi

Woah, I am glad I asked about these temps alright...

It should be noted I'm using Zalman ZM-STG1 thermal paste. It seems its pretty much AS5.
http://www.tweaknews.net/revie...ermalgrease/index3.php

I know very little of thermal grease deployment, so I took some pictures to show you guys how it was before and afterwords.

Here is the CPU when I took my cooler off:
http://www.imagebam.com/image/8b829e45569208/

The bottom of the cooler:
http://www.imagebam.com/image/c7aa3145569248/

The CPU after I applied more thermal paste:
http://www.imagebam.com/image/29734845569298/

However, placement was hard and the cooler sled on the top a bit, moving the paste around. Here is what the CPU looked like right before I finally was able to successfully place the cooler.
http://www.imagebam.com/image/60ea6445569275/

The bottom of the cooler before I was able to place it:
http://www.imagebam.com/image/a2926945569316/

Also, here is a picture of the inside of my case so you can get a feel for airflow. The wires arent pretty but I suppose it could be worse...
http://www.imagebam.com/image/e2b90b45571452


OK, here are some new temps. Please note the following factors have changed, however:
1) Now ALL fans are running at 'high' speeds.
2) Air Conditioning is on, ambient room temperature is 23C. ( Dont know prior temperature, outside it was 28)
3) I am running a slightly different Prime95 test. Last time I was doing 'blend' now I'm doing 'max FTT'. Max FTT is supposed to get highest temps..

The (high) temps are:
Core 0: 75
Core 1: 75
Core 2: 73
Core 3: 73

Once again there is a difference in temps but the gap has fallen. I'm guessing these are ok?
The temps were gained by about 35-45 minutes of Prime95. Its 1: 15 AM now so I will sleep but keep the PC on, if there is a difference in high temps I shall post them.

EDIT:
Forgot to add that the CPU is a G0 core. Voltage is auto-set at 1.2625 volts.

 

[Tempered81]

I wouldn't think that you could run a Q6600 at 3.6ghz with 1.26v... ?

 

[Reven]

To be fair that number is NOT from the bios. I left the voltage on 'auto' and the number is from the program core-temp.

 

[Tempered81]

try program called cpuz

probably say something ~ 1.46v

 

[hclarkjr]

out of curiosity what are your idle temps like?

 

[idiotekniQues]

good lord.

i run at 3.4 with my q6600 and i idle at 39 Celsius

 

[Idontcare]

Originally posted by: Reven
To be fair that number is NOT from the bios. I left the voltage on 'auto' and the number is from the program core-temp.

Coretemp reports the CPU's VID - a factory set stock voltage which the CPU should receive in order to operate at stock specs.

This is not the same as Vcore, also called Vcc, which is meant to imply the actual voltage the CPU is recieving.

To get a software-based reading of your CPU's Vcore you need to run a program like CPU-Z and watch the reported Vcore while the CPU is fully loaded (not idle).

Most BIOSes will not use the VID for Vcc when you leave the BIOS Vcore setting as "auto" and proceed to overclock the chip.

The mobo guys have a vested interest in your system seeming stable so as you overclock your CPU the BIOS has been programmed to commensurately over-volt the CPU as well unless you have explicitely selected a manual Vcore setting in the BIOS.

To run a G0 stable at 3.6GHz with small FFT in Prime95 you need something around 1.5V or higher which is why the temperature shoots so high.

 

[xj0hnx]

Originally posted by: Tempered81
I wouldn't think that you could run a Q6600 at 3.6ghz with 1.26v... ?

Mine won't even boot @ 3.0Ghz at 1.26

 

[error8]

Yup, so probably on "auto" your chip is fed by the mobo to around 1.5V. That would explain the temps.

 

[Reven]

According to CPU-Z I have a 1.400 voltage.
Low temps are:
Core 0: 40
Core 1: 39
Core 2: 37
Core 3: 38

However, it seems my PC had earlier crashed. It was fine when I left it and was working earlier this morning. However, I later found it off ( in sleep mode). Windows them prompted me " Windows has recovered from an unexpected shutdown".

 

[elconejito]

EDIT: That windows error msg wont have anything relevant. It likely crashed due to CPU instability, not because of a driver or some other issue.

Your load temps are pretty high, and your idle is pretty high too.

The first thing i would recommend is to go get some better thermal paste and reapply the heatsink. Artic Silver 5 or Tuniq TX. Apply it **according to their instructions**. More paste is NOT better. Make sure you get *all* of the old stuff off first.

==============================================
The very next thing (and this is THE most **important**) is to read the sticky in this forum about how to overclock.

I'll say it again for emphasis.... read the sticky in this forum about how to overclock.
==============================================

If you leave all of those voltages on auto and just increase the FSB you run serious risk of burning up the CPU or mobo or RAM.

After you've read the sticky (or heck, at least just skim it). If you have specific questions about what to set, then check back here and ask us. I think you'll probably find that 3.6Ghz is the upper limit of that chip and you will probably want to tone it down a little to keep the heat and voltages down. I used to own that same chip and ran it at 3.2Ghz because it would run very cool at 3.2, but required a lot more voltage (and heat) to run stably at 3.4 or 3.5 (could not get to 3.6).

 

[Schmide]

http://www.imagebam.com/image/29734845569298/


Way way way too much thermal compound. Think size of a BB or grain of rice.

 

[Tempered81]

Originally posted by: Reven


Also, here is a picture of the inside of my case so you can get a feel for airflow. The wires arent pretty but I suppose it could be worse...
http://www.imagebam.com/image/e2b90b45571452

Cpuz is reporting the actual volts in your cpu - much higher than the 1.26 you thought via bios.

You're putting too much thermal paste on there, it should just be a thin haze of grey on the HSF base & cpu.

Also do something about cable managment - like this...

http://picasaweb.google.com/lh...RGOVdQ?feat=directlink

 

[StinkyPinky]

Holy crap 98c?!? I'm surprised the cpu even runs at that temp

Try and keep it under 75 max i would say

 

[Reven]

OK, I cleaned off all the old thermal paste with some dry toilet paper and some cotton swabs. I then applied some new thermal paste, but this time, what feels like a lot less. Bellow are links with some pics I took, I'd appreciate if you guys would take a look.

I had earlier skimmed the Core 2 sticky and I now did a full read. Thats where I got the whole Prime95 idea, though I had missed the AS5 picture. I have to admit I wish I had a tube so I could replicate it, the zalman brush, while nice, makes it hard (for me at least) to judge how much I've put on.

I tried running Prime95 on the cpu again with a 1.375 voltage and 400 fsb and got a blue screen after 15 minutes. Last I had checked cores 0 and 1 were around 83 C and Cores 2 and 3 were around 67...

Anyways, at a 375 FSB and 1.36 voltage and every fan at max speeds (including the zalman) I am getting the following temperatures at idle:
Core 0: 43
Core 1: 40
Core 2: 36
Core 3: 37

I dont know the exact room temperature but the AC was on at 23.

Now I am running 4 large FTTs in Prime 95. I have run it for 30 minutes.

High Temps:
Core 0: 81
Core 1: 80
Core 2: 67
Core 3: 67

I realize 30 minutes is not enough, but its late so I need sleep. I will keep the PC on through the night

Once again it seems the 1st 2 cores are my nemesis. Considering I tightened the cooler as much as possible, I'm guessing its my thermal paste again. Unless you guys recommend otherwise tomorrow I shall clean it off again and re-apply even less. I'm assuming its still too much but when I'm applying it can feel as if I havent even covered the entire cpu yet, so I try to add more. Bad piece of intuition...


Also, I'd like to inform you guys that since I live in Istanbul, Turkey, I am unable to easily find special computer components, especially stuff like AS5. I earlier posted a link where the Zalman STG1 was reviewed along side the AS5 and it was on average only 1-2 C worse...

Thanks for all the help.

Bellow are links to some pics of the (new) thermal paste. I also tried to take a pic of the seating of the cooler just in case I had it too far forward or back, though it is hard to see.

http://www.imagebam.com/image/09dfe245669912/

http://www.imagebam.com/image/dbc33445669966

http://www.imagebam.com/image/efae2d45670003

http://www.imagebam.com/image/fa958645670042

EDIT:
I have done my best so far with the cable management, though admittedly its not much. The main problem is that I have a sata dvd drive up top and more than 3 HDDs so I have to use 2 HDD cages. On top of that 1 of the HDDs is an IDE drive. Also, the 4890 needs 2 6 pins which means an extra wire pack. I will try again though, I noticed I have some space behind the motherboard and the case and there are some holes. It will be hard getting the cables behind though.

 

[Reven]

http://www.tweaknews.net/revie...ermalgrease/index3.php

Here is that review of the zalman stg1. Now in it the guy has a picture of his thermal paste covered CPU. Isnt that too much? I seem to remember having a similar amount on my CPU...

 

[Reven]

The PC was able to stay stable 9 hours with a 375 FSB (about 3.375 ghz) without crashing. I know I need atleast 24 hours straight however I needed to shut if off and boot back with default settings for my brother.

Here were the high temperatures:
Core 0: 83
Core 1:83
Core 2: 69
Core 3: 69

I'm guessing this big of a temperature differance means that I have too much thermal paste on one side? or too little?

I noticed these temps were worse than the temps I had with a 1.4v and 400 fsb. Granted I wasnt stable for 9 hours but I had more thermal paste on last time. Maybe this time I had too little?

 

[Schmide]

Your grease seems a little thick and clumpy and regardless of what that website has. I still think it's too much. The thermal grease is just there to fill in the imperfections in the HS and chip.

You really need to control your temps before you overclock and raise your voltage willy-nilly. At that much current and those temps, your chip isn't going to last very long. Especially at 28c+ ambient. Because of that and the non-exotic cooling, I would shoot for 3.2ghz max at 1.3v or less.

Oh and edit the useless post that's screwing up the previous page. Too much information that no one cares about!!!!!!!!!!!

 

[Reven]

Alright, I've edited the post. No need to get ansy...

I dont mean to be raising the voltage 'willy nilly' on my CPU. In the sticky it says "In case you?re wondering what Intel recommends for your processor, find your chip on Intel's Processor Finder DB. The Q6600 is between 0.85 ? 1.5V. ". It also says " 1.2-1.3V - for a FSB of ~400 MHz. 1.4-1.5V ? for a FSB of ~420-440 MHz " Now, I understand that I should have turned off Auto voltage, but (emberasingly) I had trouble finding that setting. Both ways auto was running at 1.4. I then turned it lower manuelly to 1.375. The reasons I didnt lower more was because the sticky says to first get it stable with a set voltage, then lower if you can. Since I wasnt stable I didnt lower it. I'm not claiming to be an expert by any strech of the means, all I'm saying is that the above reason are why I felt the voltages I had were "OK".

Please dont take this the wrong way, I'm not trying to be confrontational and I apprieciate you trying to help, but I wanted to show my thought process.

I also dont understand what you mean about 'control your temps'. As I said earlier, I turned the AC on at 23C and kept it on. So room temperature should be around there. Room temperature should never hit 28C+, what I had earlier posted was outside temperature in the peak of summer since I had no data on room temperature...

I'll go an re-apply again tonight. I'll also try to clean up the wires.

 

[elconejito]

A Quick note on the paste... The purpose of thermal paste is ONLY to fill in microscopic cracks so that the CPU and the heatsink are in contact as much as possible. If you put too much paste, then you are actually putting a barrier between the CPU and the heatsink. Which instead of helping with temps will make it worse. So you just need to put barely enough. A very thin and even layer. The Zalman paste (IIRC) is very liquidy, so it should be easy to spread around a small drop so that it covers the CPU.

Now what you should do when you start overclocking is set the voltage and start out at stock speeds. Not set a high voltage and immediately start at the highest speed. I would start by setting your vcore to the VID (1.2626?). And setting your FSB to 266. Which is completely stock.

Run your stress tests.

*If* it's stable, then increase your FSB. I would go to 300, then 333, then 350, then 375, etc until you reach your goal.

Along the way, after you've increased your FSB, if it's not stable THEN you can increase your voltage. It will be a little trial and error as to which voltage you should increase. The sticky will help you figure out what type of error means you should "probably" increase which voltage.