Is this the right way of repasting CPU and GPU ?

Hi NBR.

Sorry for bothering you alot with my repaste questions. This will be likely my last question about repasting.



Is this the right way of repasting both CPU and GPU ?

I will be using Arctic MX-4 on the same chips in the photo ( i7-2630QM - GT 525M ).

I appreciate your help

 

That's the right way but you used way too much. do it again but only use half of the amount you have on there right now on each the CPU and GPU

 

I agree.

But it seems alright.

 

TOO MUCH!! TOO MUCH!! dude watch this

How Thermal Compound Spreads - YouTube

read as much as possible before you attempt repaste. temperatures vary a LOT.

 

Guys don't worry I haven't repasted yet
This picture is taken from L502x tear down thread.

So I should use the line method but less than the one applied in the picture ? and the dot method for the GPU but less ?

 

yes and yes. Also do a few trials (I know it takes time) to see how much spread you get with the amount of paste you use.

 

A reminder that the whole point of thermal paste is to fill in the gaps between the bottom of the heatsink/fan and the processor.
(the bottom of the heatsink is not actually flat, it has many hills and valleys)

You want CPU->heatsink as much as possible, with all the gaps filled with paste.

Paste all over motherboard is bad as is too much paste that the heatsink does not contact the CPU properly.
Being too stingy is also bad as it doesn't get enough of the gaps filled.

There are all sorts of theories as to the "best way". The truth is its somewhat of an art form and must be perfected by practice and monitoring heat. The most important part is knowning WHY you are doing what you are doing and making the best results from your knowledge and materials.

I tend to use the "five dots" method.
I put a generous dot in the middle with much smaller dots near the corners. The results are the most positive for me.

 

Yes, you are fine and have nothing to worry about. I like to spread it, but either way there is not much difference if any to worry about.

 

I have always wondered why people seem to prefer the dot method over the spread method. Doesn`t the entire chip get hot and if you have the entire chip coated with paste, the heat dissipation will be better due to better contact with the heatsink than compared with the dot method where you usually get corners and outer edges of the chip without paste?

The five dots method Kernal speaks sounds good though since there are 5 blobs that will spread all over the chip evenly

 

Glad that the shape of the die for the GTX680M is back to square, I hated GTX580M/675M's rectangle die .

 

Yeah its good but I`m still sceptical that you will get the blob to spread out all over the chip even with a square

 

The PK-1 paste blob was around 3.5mm and it as enough to cover the whole die on the GTX680M. On the GTX675M 5mm blob couldn't cover the whole die, had to do horizontal line method.

 

Pics or it didn`t happen

 

The video up above shows how the paste spreads fairly well. It spreads in a circular pattern pretty much no matter how you place it.
Too much paste results in either spillage over the edge or excess paste that stops or reduces cpu->heatsink contact that is better than cpu->paste->heatsink contact.

The 5-dots method takes some practice, but results in fairly good coverage and since it is a small amount of paste near the edges, usually doesn't spill over. Too much is still too much though.

The single-dot method is least likely to spill over, but does indeed leave the edges of the cpu heat spreader without paste more often than not.

The "spread" version takes a lot more perfection IMHO as the excess paste has nowhere to spread and often results in air-bubbles that the "dot" methods force out by displacement. It also tend to be messy as you need a way to spread it thin enough to not be too much, but thick enough to fill the gaps. Oddly enough the dot methods and the displacement they cause tends to naturally fill gaps and push the excess to the edges, but with just a little practice tend to be thin enough to not cause spills.

While it IS good to get the best coverage you can, coating the whole area of the outside of the CPU isn't necessary, it just improves the system as it increases the surface area of better heat transfer.

The problem is, CPU->paste->heat sink surface area is not nearly as good as CPU->heatsink surface area and that people are far from perfect in getting everything covered properly. Furthurmore, air bubbles usually caused by "spreading" paste tend to counteract any benefit you get by using the spread method unless you are much more perfect at it than most people.

Finally, paste is messy and tends to get all over everything. The less you touch it the less likely you are to get it all over something its not supposed to be on. The "spread" method tends to make bigger and more destructive messes IMHO.

Again, its all personal taste and the end result is more important than the method. However, the above is why often people who "repaste" end up with worse heat than they started with despite the better materials.

 

I haven`t read anything yet of the responses or the posted the video, but this is how I visualize how the paste is gonna spread out with dot method. And as you can see, corners without paste. Bigger blob to cover the corners, means paste will be running outside the chip for the paste that met the edge with the first attempt

 

Next week when I receive my GTX680M from RMA I can post that

 

You make good points. I guess it comes down to what is the best scenario: Air bubbles here and there with the spread method or areas not touched with the blob method? What gives the best temperatures. CPU with heat spreader? I thought mobile CPUs came without?

I just watched the video posted on the first page. There do seems to be quite a bit of air bubbles with the spread method. Arctic didn`t seem to spread all over the chip with dot method either, neither did the silicone. I guess I have to take Hackness word for that its possible, but I still have doubts that he can do it without making a mess.

I`ll look and see if I can find any reviews that have tested them out

Finally getting your 680M back ey? YAY. I just recieved my next GPU in my mail today. Will install it when I get my MSI
Looking forward to some pictures of the paste job

 

Now I'm extremely confused , I have to dismantle the whole laptop to get the heatsink so I can't just repaste then measure the heat then repaste again, I was searching for the best technique because I prefer not to open the laptop alot ( I hope this is going to be the last time this year ).

What should I use.. Line method , spread method , three dots method or what ? :S

 

50 dot method

 

If this method guarantee the best coverage, I'll use it

 

Benchmarksreviews say the "two line method" is the best. Or the spread method. Or with a bit of scepticism, the cross method
Benchmark Reviews: Performance Computer Hardware Tests

 

Hardwaresecrets says the best method is the dot method. They even included temperature measurements to compare. Yeah, so there you have it. Go with the dot method.

What is the Best Way to Apply Thermal Grease? - Part 1 | Hardware Secrets

 

There's a pretty popular youtube video that compares various repaste methods, may be of some use to you.


Apologies if it was already posted above.

 

@OP, that is perfectly fine. Maybe a little too much but not bad at all. Won't hurt anything.

 

I think I will go with either the three dots method or the small line method.
Thanks guys for your help.

+rep for most of you ( 24 hours ).

 

@OP, that is perfectly fine. Maybe a little too much but not bad at all. Won't hurt anything.

Spread method usually results in air bubbles. X method seems to have best results for me with IC diamond. Just note that on a laptop CPU/GPU you need FULL COVERAGE because there is no Integrated Heat Spreader (IHS) like on a desktop CPU. The entire die is exposed so it needs to be fully covered. That is why it's better to have a little more than not enough.