PC overclocking question

[Kawazu]

@Kawazu

Do I REALLY need to remove the heat spreader ? Looks risky is all I mean.

Or would it be enough if I remove the heat sink again, clean the heat sink and CPU / spreader and only apply a small amount ("pea sized") in the middle of the CPU / Spreader ?

I'd like to get this right first time now.

No no no, jusy wanted to show you what it looked like under it so you understand why you dont need to cover the whole thing with thermalpast.

Just put a smal pea on the cpu and let the preasure of the heatsink even it out when you get it in place.

 

[dougal]

OK, so I re did the thermal paste... bought Arctic 4 (5 out of stock).

Removed, cleaned (both CPU and heat sink) and then applied pea of paste on cpu centre.

Assembled, started with the cpu running at 2924.85Mhz and ran Prime95 for 35 minutes.

Max temp on Core#0 and Core#1 (using core temp) was 73 degrees
Max temp on Core #1 and Core#3 was 69 degrees

The CPU cores are currently running between 49-52 degrees now. (Obviously without Prime95 running)

In the whole 35 minutes it did not crash and passed the tests successfully.

Bear in mind Malta is really hot and the room where I keep the PC in is probably the hottest room in the house (top floor and surrounded by windows). Also bear in mind that Prime95 pushes the CPU to 100% usage.

Is the CPU OK like that ?
Can I push it a bit more ? (Say 3.2Ghz?)

 

[Kawazu]

OK, so I re did the thermal paste... bought Arctic 4 (5 out of stock).

Removed, cleaned (both CPU and heat sink) and then applied pea of paste on cpu centre.

Assembled, started with the cpu running at 2924.85Mhz and ran Prime95 for 35 minutes.

Max temp on Core#0 and Core#1 (using core temp) was 73 degrees
Max temp on Core #1 and Core#3 was 69 degrees

The CPU cores are currently running between 49-52 degrees now. (Obviously without Prime95 running)

In the whole 35 minutes it did not crash and passed the tests successfully.

Bear in mind Malta is really hot and the room where I keep the PC in is probably the hottest room in the house (top floor and surrounded by windows). Also bear in mind that Prime95 pushes the CPU to 100% usage.

Is the CPU OK like that ?
Can I push it a bit more ? (Say 3.2Ghz?)

Hard to say, you get one temperature when you use prime95 and another when you play games. If you get 75-80c in prime95 you will probably not get over 70c when you play games.

If you look at intel´s specs on the CPU it say The B3 is 62.2C and the G0 is 71C
so try keeping the cpu under 71c when you are playing.

Place the computer in a coole room or get even better CPU cooler, will be hard to cool down the cpu if the room temperature is high

 

[dougal]

There are 2 case fans. One in front and one at the back. They are not very fast.

Should I get another 1 or 2 fans ?

Could I drop the voltage and raise the FSB to hit 3.2 safely ?

Sorry for all the noob questions but I really appreciate your help

 

[Uncle_Meat]

I wouldn't even go as far as pea sized, more "grain of rice" kinda thing, maybe a bit more nicely and *thinly* spread out.

I'm running a Q8300 OC'ed to 3.2Ghz with a Freezer Pro 7 and I only just manage to tip the temperature scales at 45 or so C on that full tilt. Saying this, when I was temp-testing it was winter in an unheated workshop, so ambient was single digit temps.

Best I've seen in here, couple of winters ago when we were really in the freezer, firing up a PC and checking temps in BIOS. CPU was 5C and ambient was -3

 

[Kawazu]

There are 2 case fans. One in front and one at the back. They are not very fast.

Should I get another 1 or 2 fans ?

Could I drop the voltage and raise the FSB to hit 3.2 safely ?

Sorry for all the noob questions but I really appreciate your help

To know if its worth getting more case fans or not is simple. Remove the side of the computer if the temperatures drop alot you have to litle airflow running thru your case. Then add more fans. If the temperature stays about the same there is no need to add more case fans.

If i where you i would drop the voltage and see how much i can drop it before the system becomes unstable.
Remember, only change one thing at the time, or else it will take forever to know what is causing the instability.

This is how i would do it.

1: Lower the CPU multiplier
2: Raise the FSB
3: Boot the system and check if its stable if it is go to 2: again and continue untill its not stable any more, then you lower the fsb by say 5-10mhz.

4: Keep the FSB at the stable point you found in 3:
5: Raise the CPU multiplier by 1 step at the time.
6: Boot computer and see if its stable if it is, rise it 1 more.
7: if its not stable try to increse the voltage of the CPU by smal steps. and continue to do so untill its stable.

Stay under 1.5V and keep an eye on the temperature.

As you see you have to start increasing one part at the time to be sure that you get a stable system at the lowest possible voltage. Takes time but its worth it in the end.

---------- Post added at 14:08 ---------- Previous post was at 13:57 ----------

I wouldn't even go as far as pea sized, more "grain of rice" kinda thing, maybe a bit more nicely and *thinly* spread out.

I'm running a Q8300 OC'ed to 3.2Ghz with a Freezer Pro 7 and I only just manage to tip the temperature scales at 45 or so C on that full tilt. Saying this, when I was temp-testing it was winter in an unheated workshop, so ambient was single digit temps.

Best I've seen in here, couple of winters ago when we were really in the freezer, firing up a PC and checking temps in BIOS. CPU was 5C and ambient was -3

The amount of thermalpaste you use is not that important and long as there is somthing there between the two parts, using to much of for example arctic silver aint good just becouse it can short stuff that it comes in contact with.

Just look at this, they are using toothpaste and the difference in temperature is nothing to fight about

http://www.hardwaresecrets.com/printpage/Thermal-Compound-Roundup-July-2011/1334

 

[Uncle_Meat]

The amount of thermalpaste you use is not that important and long as there is somthing there between the two parts, using to much of for example arctic silver aint good just becouse it can short stuff that it comes in contact with.

Just look at this, they are using toothpaste and the difference in temperature is nothing to fight about

http://www.hardwaresecrets.com/printpage/Thermal-Compound-Roundup-July-2011/1334

I would say too much is worse, I've fixed overheating PCs on many an occasion by cleaning off the mountain of gloop that someone has put on there and reapplying a sensibly meagre amount.

As for the actual compound, I agree. I've been using cheap Servisol thermal compound, the stuff that comes in a toothpaste sized tube for a couple of quid a go, for years and have not yet noticed any ZOMG! difference between that and the more expensive stuff.

 

[Kawazu]

The amount of thermalpaste you use is not that important and long as there is somthing there between the two parts, using to much of for example arctic silver aint good just becouse it can short stuff that it comes in contact with.

Just look at this, they are using toothpaste and the difference in temperature is nothing to fight about

http://www.hardwaresecrets.com/printpage/Thermal-Compound-Roundup-July-2011/1334

I would say too much is worse, I've fixed overheating PCs on many an occasion by cleaning off the mountain of gloop that someone has put on there and reapplying a sensibly meagre amount.

As for the actual compound, I agree. I've been using cheap Servisol thermal compound, the stuff that comes in a toothpaste sized tube for a couple of quid a go, for years and have not yet noticed any ZOMG! difference between that and the more expensive stuff.

That dosent make sence, even if you buy the cheapest paste there is it still have a better heat transfer the air.
And if you place to much on the cpu the weight and the clips will force the thermal paste out from the core so the layer would be exactly the same as if you put on less.
The only difference would be that the core is now isolated with thermal compund wich still have a better heat transfer then air.

 

[Uncle_Meat]

That dosent make sence, even if you buy the cheapest paste there is it still have a better heat transfer the air.
And if you place to much on the cpu the weight and the clips will force the thermal paste out from the core so the layer would be exactly the same as if you put on less.
The only difference would be that the core is now isolated with thermal compund wich still have a better heat transfer then air.

I agree with the paste debate. Cheap paste is better than air, obviously, and I've not yet found any compelling reason to use the expensive stuff over the Servisol stuff that I've used for years.

As for the HS and clips spreading it out, I can only go on what I've found and seen over the years, where upon removing the HS from the CPU there's been quite a thick, dried out almost chalky layer of paste which in some cases I've had to scrape off with a razor blade before washing it in IPA. Reapply paste and temps have dropped.

With some consistencies of pastes that I've seen though, I'm not convinced that the pressure applied by the HS on the CPU would be enough to squeeze it *all* out to a fine layer and also, as you say, some paste is conductive and having that spread out here, there and everywhere is definately a Bad Thing.

OK, some of this can be the pre-applied stuff they use on some., but seen quite a few home builds where they've gone overboard with the application and it's dried out to quite a thick layer, which is no good.

Sony laptops for example. I've had quite a few of thesee overheating themselves to shutdown due to a thick layer of paste. You could run Prime95 along with realtemp, for example, and watch the CPU heat up 60-70-80-90-SHUTDOWN! in seconds. If you touched the heatsink, it would be nowhere near as warm, conclusion being that in these cases the paste is acting more as an insulator. Remove, clean, reapply and things were all back to normal, the same test nowhere near as hot and, just as importantly, the temps have dropped like a stone again once the Prime95 task has been stopped, proving we have thermal transport again.

In an ideal world with ideal manufacturing, the HS and CPU would be perfectly flat with no imperfections and you wouldn't need paste at all, metal to metal being the best conductor of all. However, unless you're going to hand lap all the mating surfaces, paste it is and only just enough to compensate for those slight imperfections in the mating surfaces, just enough to ensure no tiny air pockets.

 

[Kawazu]

Here is a test where they have used different forms of application.
http://www.hardwaresecrets.com/printpage/What-is-the-Best-Way-to-Apply-Thermal-Grease-Part-1/1303

The sony computer you are talking about are most surley using somr sort of thermal pads like they did on my Asus laptop, was alot of it pushed out of thr sides and was like concreat so i removed this and add some AS5 and thr temp droped 4c not much but the stuff they use from factory is utterly ****

 

[johnim]

here is a video i normally just put a thin line down the middle of the cpu and use mx2 or mx4

http://www.youtube.com/watch?feature=player_embedded&v=EyXLu1Ms-q4