Take thermal paste 'burn in' into account when Overclocking?

[Random_guy]

When you are overclocking, do any of you take into account thermal paste 'burn in' whereby the thermal paste inproves in efficiency with time?

A bit over a year ago I decided to overclock my CPU, bought a new cooler, read several guides, and successfully managed a mildly respectable 3.2GHz on my E6750 (2.66GHz at stock).

Before I started, I stress tested the CPU at stock with the stock cooler, and freaked out a bit when it sailed past 75c after about 45 mins!

After putting the new cooler on with the shiney new thermal paste, the temps dropped to just under 65c on stock speeds after a couple of hours stress testing.

I then overclocked the processor (including a drop in voltage to try and lower the temps a bit), and ended up with the temps being just under 65c after an overnight stress test.

Flash forward a year or so, and I'm getting ready to switch out my E6750 for my ebay-tastic Q9400 to tide me over until Sandy Bridge has matured (possibly even until Ivy bridge has come out). I decided to baseline the temps and found to my amazment that after several hours stress testing, the temps are staying south of 50c. In fact, the total increase in temps from idle to load is <10c.

Ambient temps are roughly the same (the heating is on in the room, so the temperature is roughly what it would have been when I first overclocked), so the only thing I can think of is that this 15-20c drop in temps is due to the thermal paste burning in.

I had no idea there could be such a large effect over time. If I had known, I might have revisited the overclock after a little while and shot for a higher clock speed.

So do you any of you take this into account when you stress test a new build?

[Uriel]

Depends on your compound. Artic Silver improves with physical realignment of metal particles, which takes time. Arctic MX-2 and 3 do not and theoretically work instantly.

[pipTheGeek]

I always thought the effectiveness of the thermal grease would decrease over time as it dried out. How odd.

[redddraggon]

I always thought the effectiveness of the thermal grease would decrease over time as it dried out. How odd.

Well it pretty much depends on the liquid used in the thermal paste. A solid metallic layer will conduct the heat far better than the liquid used.

[Pob255]

As has been said it depends on the paste used and how it reacts over time.

Personally I don't, but do check temps a couple of months down the line to make sure nothing is going wrong.

Think of it this way, if your paste improves over time and your temps go down, then that's not a bad thing, so if you're in a safe range when you first overclock, then a couple of months later your temps will be even better, temps below max are better than temps above max or even close to max (as a coolers effecency can go down over time as dust build up on it.)

[lego batman]

Hate to ask this but I will just because the improvement of 15 degree is so huge - is this normal behaviour for thermal paste? I certainly haven't noticed such a temperature drop which makes me think I might have applied it incorrectly. But in that case, why doesn't MX2 have that sort of temperature advantage over newly-applied Arctic Silver? Which leads me back to the first question - anyone else had such marked improvements?

[redddraggon]

OP - Are you sure the CPU is still OCed and the CMOS hasn't been reset?

[Random_guy]

Yep, its still overclocked, at least according to CPUZ! I just dug the thermal paste out, and its Arctic Silver 5, and does say that temps will drop several degrees over the break in period (~200 hours for AS5).

Pob255: Good point about dust build up on the cooler, hadn't thought of that. I was just annoyed, as I could have gotten a bit more performance out of the CPU, but I hadn't wanted to push it any further because of the temps (stability was fine, given that I was able to overclock and undervolt at the same time without it skipping a beat).

I will definitely be revisiting the new CPU around xmas time to see what the temps are like, and depending on stability, possibly seeing if it'll clock higher (this is assuming that it overclocks ok in the first place *fingers crossed*).

Incidentally, when changing CPUs on an overclocked board, will the BIOS generally detect a new CPU and revert to stock settings, or will it keep the higher FSB, and I should therefore change the settings back to stock before starting the upgrade?

[Pob255]

Best to rest any cpu settings to Auto then re-do the overclocking
Different motherboards will react to a cpu change in different ways.

[ERU]

I still use the tube of 'Arctic Silver 3' paste i've had for 5+ years. Based on what has been said here am I still safe to be using it? It seems to be fine ... Would buying something newer improve my temps? Is my CPU in danger of death?

[GoNz0]

I still use the tube of 'Arctic Silver 3' paste i've had for 5+ years. Based on what has been said here am I still safe to be using it? It seems to be fine ... Would buying something newer improve my temps? Is my CPU in danger of death?

it should be fine so long as it not separating in the tube, ie clear liquid coming out and thick paste after.

and 15 degrees is just about impossible for thermal paste over a year, some other factor is at work there.

[Pob255]

Yes thinking about it 15 degrees seems far too large a change, 2-3 degrees would sound right but not 15
I'd guess other changes to cooling setup or PWM fan settings.

[Random_guy]

Yes thinking about it 15 degrees seems far too large a change, 2-3 degrees would sound right but not 15
I'd guess other changes to cooling setup or PWM fan settings.

I hadn't changed anything that I can think of. The only thing I can think of is that I was using a later version of the two programmes I was using to monitor temps (Speedfan & Core temp) and they might have changed their minds about what TJMax is for my CPU, and so gave me lower temps this time round?

Anyway, I've made a note of the temps I'm getting with my new CPU, so I'll make a memo to check again around Xmas time to see if there's any difference.