Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
@fr4c
Awesome results Roman, I will copy/paste your results and place it on the main page. Good to see more members taking the initiative to document their procedures.
I guess it was my way of giving something back to the community :). My pleasure. And thanks for including my post in the main post!

@aznguyen316
Roman! Glad you got around to doing it yourself. It always feels better doing it, plus you'll learn a thing or two eh?
Most definitely. It feels so good to know exactly how all this hardware is put together in there. Also, it brought me peace of mind as there's no more doubt as to whether things are clean or butchered (a founded doubt, as it turned out :rolleyes:).

Glad my video was of help, I really appreciate the shout out haha.
To say it was of help is an understatement. It's the reason I felt confident going ahead with the repasting and also why I have been successful at it. The shout out was the least I could do, well deserved ;).

Anyway, I'm dissappointed at what Apple's repaste job. Pretty shoddy and terrible!
Same here. Though the more I think about it the more I feel I can't blame them. I can't imagine an automated robot that would apply just the right amount of paste while ensuring it would spread evenly. Even less so for a factory worker who's only in line to pay the bills. It's kind of a minute job that would require too much work per machine for Apple to bother.

I didn't even bother with teh southbridge but that's kind of interesting. Great pictures. So on the southbridge and thunderbolt, was that stock thermal pads you used or did you apply those yourselves. At least they look like pads?
The stock weren't thermal pads but rather a thick white thermal paste (as opposed to the gray one on the CPU and GPU). I replaced this filth with 1mm or 1.5mm thick thermal pads, leftover from an EK waterblock. They proved to be a perfect fit.

Your results closely mirror what my results were. 88C is basically my average high, 90C for a second before fans kick in all the way and never higher than 88C. From there it only gets cooler as the fans keep running on blast. I played more SC2 tonight and it goes up ti 86C-88C and then after running at 6000RPM it'll cool it to 80C or so and run at 5500RPM.
Nice. That's the temperature + fan-speed behaviour that feels right, namely what is intended by Apple for these new Sandy Bridge MacBook Pros.
 
I have gone ahead and re-applied the thermal paste on the 4 chips (CPU, GPU, Thunderbolt controller, southbridge).

Before:

GPU heatsink:



Dirty. Nice job, Apple "Geniuses"... (I had them re-apply thermal paste on both the CPU and GPU.)

Southbridge heatsink:



Excess of thermal paste coming out of the edge of the heatsink. Originally, I wasn't planning on re-doing the southbridge or the Thunderbolt controller, but upon seeing such horror, I would'nt have been able to live with myself knowing that such a mess was living underneath my MBP's keyboard.

GPU & CPU:



Nasty. They told me they cleaned up the stock application and re-applied only the necessary amount. Right :rolleyes:. Not to mention, the screws were only loosely tightened. They did swap the motherboard for a new one, covered by the first year Apple Care (in order to alleviate my doubt about the CPU coming from a poor batch), so I can't complain about them specifically.

Thunderbolt controller:



Kind of clean, relative to its neighbours, though way too thick and uneven.

Southbridge:

I'm sorry I forgot to take a picture of the southbridge, but it wasn't pretty, as the picture of the corresponding heatsink (above) can attest.

After:

Thunderbolt controller & southbridge:



As documented by somebody early in this thread, their heatsinks are elevated by a good millimeter above the corresponding chips. Thermal paste isn't appropriate in this case. I put thermal pads leftover from an EK waterblock. They fit perfectly underneath the heatsinks: there's a good, tight contact after screwing them back on.

GPU & CPU:



All cleaned up. I found a toothpick to be the perfect tool for removing thermal paste between the tiny elements (transistors?) surrounding the main chip. Precise yet soft wood.

GPU & CPU:



Arctic Cooling MX-2 applied, ready to be spread by the heatsinks. There's a little more than necessary but at least I'm almost certain it spread over the whole surface. Also, the MX-2 is liquid enough that the pressure from the heatsink only leaves the thinnest layer necessary to fill surface imperfections while excess is pushed to the edges (proved over and over after dozens of desktop CPU and GPU heatsink and waterblock un-/mounting).

Results (running on battery, integrated GPU)

Absolute maximum load temperature:
93°C => 88°C
Note: temperature rises to 94°C but only briefly as when when fans reach 6200 RPM, the CPU is cooled town to a stable 88°C. Before, temperature rose to 93-95°C and stayed there.

Idle temperature (no applications loaded, Finder only):
45°C => 37°C

1080P video in VLC:
85°C => ? (not tested yet)

1080P video on YouTube:
85°C => 67°C
Brief initial peak at 76°C tightly controlled by fans speeding up to 2500 RPM for less than a minute.

Average usage (multiple Chrome tabs, paused YouTube videos, iTunes playing, MPlayerX paused):
55-65°C => 45-47°C.

General:
Before the "re-pasting", the top left area of the keyboard was always hot (but still sustainable to the touch). YouTube videos would make it hot the point I couldn't leave my finger on it. Fans rarely spinned at their minimum speed (2000 RPM), and it felt like they were useless.

Fans now stay at 2000 RPM nearly all the time. They did speed up to ˜2500 RPM (barely audible in a silent room) once when the CPU cores reached 85°C until they got it down to 76°C, and proceeded to progressively slow down to 2000 RPM (their minimum, inaudible). I love this behaviour.

The top left area of the keyboard rarely gets hot anymore. Warm sometimes. Slightly warm most of the time. That's because the CPU never gets to stay hot long enough now: fans actually manage to move heat away from the heatsink now.

To me, that was the whole point of this operation. Temperatures aren't that much lower, but the top left area is now cool and I never hear the fans. Exactly what I expected.

Difficulty:

Very easy, actually! aznguyen316's teardown video guide on YouTube was the exact and only instructions I needed to proceed. This video is brilliant, really! Much thanks again to him for this gem, a gift to other MBP 15 2011 owners.

Tools needed:
  • Phillips screwdriver
  • TX6 Torx screwdriver
  • Toothpick (in place of spudger): for both unplugging the various kinds of cables, and cleaning up intricate parts of the chips
  • Nailpolish remover
  • Coffee filters
Conclusion:

Success! Thrilled with the result. Smooth, easy process (for the most part thanks to aznguyen316's video guide). 100% clean chips. I can only recommend thermal paste clean re-application to all owners of an MBP.

As an added bonus, I now know my MBP inside out. It's not a black box anymore. Apple produce some serious quality products, zero doubts about that. The tight internals just speak for themselves. Sturdy components, neat cable routing, sleek black PCB motherboard with robustly soldered elements. Eye-candy for me :). It's a shame that the assembly chain ends with such a poor job at applying thermal paste. Neglected as seemingly irrelevant, but absolutely vital in the end.

Great job! You have applied the thermal paste on Everything! ;)
 
My experience

Hi everyone.

First, big thanks for this great post.

After replacing 3 (!!!) macbook pros (first had light leakage, second had bad quality, unpolished aluminum. Thank you Amazon!!! Amazing service!), My third one seemed to be good, however it also seemed to have heated up more than the previous ones.

I opened the case using the aznguyen316 excellent video from the first page post (thanks!). I actually have a 17" Macbook Pro and not 15", but they were similar enough to figure out. Just make sure you are gentle.

Anyway... I had a zoo of thermal paste over there. See below. I cleaned it up with arctic cleaner 1 & 2, then applied Zalman Super Thermal Grease ZM-ZTG1.

The process was ok - not as scary as I thought. However, I now still notice considerable fan noise and temperature, so I might have been doing something wrong. Can anyone give me any ideas?

Temperatures are about 87-90 degrees when encoding the handbrake test clip (see https://forums.macrumors.com/threads/1111126/) , fans at 5500rpm. Beforehand it was between 81-86 degrees, and fans were at 4200rpm.

EDIT: At Idle, my CPU is now at 54 degrees. GPU at 50. How are yours?
 

Attachments

  • DSC_0047.jpg
    DSC_0047.jpg
    176.2 KB · Views: 381
  • DSC_0049.jpg
    DSC_0049.jpg
    139.1 KB · Views: 118
  • DSC_0050.jpg
    DSC_0050.jpg
    105.8 KB · Views: 121
  • DSC_0053.jpg
    DSC_0053.jpg
    148.2 KB · Views: 135
  • DSC_0054.jpg
    DSC_0054.jpg
    116.7 KB · Views: 137
Last edited:
My guess would be that you applied it too thickly. According to most of the directions I saw, it gets applied to the processors (GPU and CPU) in a thin coat and then it is just used as a prep coat to prep the heat sinks (put on and then wipe off just leaving a haze and not an actual coating) on the heat sinks.
 
odedia, the original paste looks pretty bad. I'm surprised your temps would go up from that. Give it a few more tests and general usage and see if you notice any better fan/temp behavior.
 
@odedia
Thanks for the pictures and report.
My guess would be that you applied it too thickly. According to most of the directions I saw, it gets applied to the processors (GPU and CPU) in a thin coat and then it is just used as a prep coat to prep the heat sinks (put on and then wipe off just leaving a haze and not an actual coating) on the heat sinks.
+1. You shouldn't have applied thermal paste onto the heatsink.
Also, have you tightened the screws all the way?
 
@odedia
Thanks for the pictures and report.

+1. You shouldn't have applied thermal paste onto the heatsink.
Also, have you tightened the screws all the way?

Hi Roman,

I tightened the screws as much as possible, didn't want to risk breaking something. 1 screw is a little on the flaky side.

I re-applied the thermal paste as suggested. I used a very very thin layer on the CPU/GPU only, and nothing on the heatsink. I again used arctic cleaner 1&2 plus Zalman ZM-STG1. I used a hair-dryer to dry the cleaning residue before applying the thermal paste. There was a little bit of lint on the cotton balls I used the first time, so this time I tried something else. I don't have any coffee filters, but you'll be surprised how useful "female" pads can be at this situation :).

I tightened the screws as hard as I could without breaking something.

Booted. All seems well.

Ran my sample test once again (big buck bunny handbrake encode).

This time, the CPU spikes to as high as 94 degrees :(. Fans are at their maximum. GPU is not as hot, but it is not really used in the encoding.

At idle (as in, now, while typing, chrome is the only thing running), the CPU is at 57 degrees and the GPU is at 53.

I attached the iStatPro readings from when I ran the stress test. Is it normal for the heatsink of the CPU to be so low while the CPU itself is so high? What am I missing here? I wonder if the thermal compound I am using is the reason now? Has anyone had success with it?
 

Attachments

  • Screen shot 2011-04-19 at 10.45.28 PM.png
    Screen shot 2011-04-19 at 10.45.28 PM.png
    78.7 KB · Views: 175
@odedia
OK. These temps don't seem normal indeed. I have done the test (Big Buck Bunny, Apple TV 2 -- on battery, integrated GPU). The CPU reached a maximum of 84°C. The fans settled on 3000-3100 RPM, lowering the CPU temperature to a stable 80-81°C. What's interesting is the heatsink is at a stable 56°C while yours is at 50°C: this is (kind of) good news for you, it means there's room for improvement.

Before thermal paste cleanup / re-application, HandBrake would make temperature reach 93°C with fans at 6200 RPM.

As you may have read in my post above, I have applied Arctic Cooling MX-2 without spreading it myself. The heatsink did that on its own. I am almost certain that is what you need to do to ensure a proper contact.
 
@odedia
OK. These temps don't seem normal indeed. I have done the test (Big Buck Bunny, Apple TV 2 -- on battery, integrated GPU). The CPU reached a maximum of 84°C. The fans settled on 3000-3100 RPM, lowering the CPU temperature to a stable 80-81°C. What's interesting is the heatsink is at a stable 56°C while yours is at 50°C: this is (kind of) good news for you, it means there's room for improvement.

Before thermal paste cleanup / re-application, HandBrake would make temperature reach 93°C with fans at 6200 RPM.

As you may have read in my post above, I have applied Arctic Cooling MX-2 without spreading it myself. The heatsink did that on its own. I am almost certain that is what you need to do to ensure a proper contact.

Well, I guess tomorrow is another day then... I'll get some other thermal paste and retry.

Any reason to prefer MX-4 over MX-2? Is it supposed to be a better version?

Are you referring to the "Grain of rice" method for application? And how tight have the heat sink screws? I am worried about breaking something there, it feels very fragile.
 
Well, I guess tomorrow is another day then... I'll get some other thermal paste and retry.

Any reason to prefer MX-4 over MX-2? Is it supposed to be a better version?

Are you referring to the "Grain of rice" method for application? And how tight have the heat sink screws? I am worried about breaking something there, it feels very fragile.
Heat sink should be tight, but not overly tight as to strip the board. Once the screwdriver stops turning, torque it just a bit and you should be fine.

What is the ambient temperature that you're using the MBP in? The temps, especially that of idle, seems a bit high with just Chrome running.
 
For those who are thinking about doing this "upgrade", I'd keep a few things in mind:

1) The heatsinks and heatpipe designed in the macbooks are used to wick away the heat from the cpu/gpu - The heat sink is only so big since it has to fit into the mbp's casing. Therefore it can only help so much.

2) The thermal paste isn't going to make your temperatures any more different than stock thermal paste. The thickness of the stock setup is indeed way too much to be efficient.

3) You will for sure see significant and quicker cpu temperatures FALL when the CPU is going from full load to idle. It helps substantially if the paste applied is thin enough to allow the heatsinks to do their jobs properly.

You won't see your full load temperatures any more different than everybody else's since the heatsinks are identical. So don't expect to see 70C temperatures when running all 4 cores and all 4 virtual threads being taxed!

82-92C is average for my MBP 17" 2011 after reapplication. That's with the fan up to 5600rpm to bring temperatures back down to about 83-85C.

But the CPU temperatures do FALL substantially faster back to around 45C or lower depending on ambient temperatures when the CPU goes from full load back down to idle.
 
@Roman.

I let the machine stay put, shut down for the night.

Booted it up this morning, things seemed quite cool.

Ran the famous test, this time - I did it exactly like you did: On battery, integrated graphics. Before this, I was testing plugged in, using discrete graphics and connected to an external full HD monitor as primary display, laptop monitor as secondary display.

Using your setup, things are cooler. See results below, as you can see - they are very similar to yours. So I am a bit more relaxed now :).

Do you think I would benefit from replacing the application with an MX-2 at this point?

By the way, GPU is now very cool. I can't get it anywhere near 70 degrees, even with GPU benchmarking apps.

Last attachment below is of the current, safari-only idle mode. Feels good :).

EDIT: Playback of youtube full HD video using discrete graphics (This one http://www.youtube.com/watch?v=-oCCnxBos10) gives the following readings:
CPU 69, CPU heatsink 52, GPU 60, GPU heatsink 61, rest are in their 30's.

EDIT #2: I really don't know what's going on. Now just having the laptop sitting there, with nothing open, not even Safari, CPU got to 64 degrees idle. I'll give it another day or two and see what happens.
 

Attachments

  • Screen shot 2011-04-20 at 7.25.59 AM.png
    Screen shot 2011-04-20 at 7.25.59 AM.png
    74.7 KB · Views: 140
  • Screen shot 2011-04-20 at 7.26.47 AM.png
    Screen shot 2011-04-20 at 7.26.47 AM.png
    68.9 KB · Views: 143
  • Screen shot 2011-04-20 at 7.28.27 AM.png
    Screen shot 2011-04-20 at 7.28.27 AM.png
    70.3 KB · Views: 116
  • Screen shot 2011-04-20 at 8.19.25 AM.png
    Screen shot 2011-04-20 at 8.19.25 AM.png
    52 KB · Views: 111
Last edited:
Successfully did mine.

in handbrake, encoding sits at a stable 80C-85C @ 4200rpm fan speed

idle with nothing on with Integrated graphics = 38C.

idle with nothing on with discrete graphics = 43C.

I'm well pleased.
 
Last edited:
Wow, this was worth doing, thanks! Idlie temps on Intel graphics 35-45 degrees, 40-50 on radeon clamshell closed with my 24" LCD. Drops of 5-10 degrees all round, and fans don't spin up so much. Thanks to aznguyen316 for the video, only resource I ended up using. I found it easier with the Magsafe connector removed, it's easy to remove from the motherboard side once you have it out of the unibody.

Here's the damage before:
 

Attachments

  • photo.JPG
    photo.JPG
    571.2 KB · Views: 147
I don't think there are metal caps over the core on the newer MBP processors, so the spread method would be better to make sure the entire surface of the core touches the heat sink. Correct me if I'm wrong though. The pea method or rice grain method would still leave the corners uncovered by thermal paste.

However... the spread method does seeem to introduce air bubbles... a nightmare in this situation.

I think when I replace my thermal paste I will probably use the rice grain method for the fear of bubbles but does anyone have an "expert" opinion on this?

Oh and I have trouble finding a Torx screwdriver over seas, quite annoying.
 
Last edited:
Regarding thermal paste application techniques, everybody has their opinions, but yaochoon on YouTube has made some interesting videos showing how each technique translates into surface coverage and creation of air bubbles:
He also posted his theory on why the manual spread technique is probably not the best (and I fully agree with it):

[url=http://img716.imageshack.us/img716/9131/cbf792db.th.jpg]Image[/URL]

There are different recommended techniques depending on the type of CPU. There is no one size fits all.

From the horse's mouth:
http://www.arcticsilver.com/intel_application_method.html#
http://www.arcticsilver.com/methods.html
 
Well, I guess tomorrow is another day then... I'll get some other thermal paste and retry.

Any reason to prefer MX-4 over MX-2? Is it supposed to be a better version?

Are you referring to the "Grain of rice" method for application? And how tight have the heat sink screws? I am worried about breaking something there, it feels very fragile.

MX-4 is newer than MX-2 and supposedly shaves another degree or two from the idle temps. I used it for my application and I'm very happy w/ the results.
 
MX-4 is newer than MX-2 and supposedly shaves another degree or two from the idle temps. I used it for my application and I'm very happy w/ the results.

Yeap. The MX-4 is a lot easier to apply than MX-3. What had happened was that while MX-2 viscous, it's successor MX-3 was not. Thus they came out with MX-4 which had the thermal improvements of MX-3 while bringing back the viscosity of MX-2.
 
need to remember,it takes months for the paste to settle down,it has to go thru heating and cooling cycles before it works at optimum .

great job,i did mine,hideous too,way too much on it like in the pics.
 
@aznguyen316
I was watching ur youtube video and in the video you mentioned don't use the Arcticlean 2 on the CPU/GPU, why?

I'm going to redo the thermal paste once I got my Macbook Pro next month :) and thanks for the excellent video guide.
 
Oh I figured because it's a thermal prep I felt better just using on the copper heatsink surface and wiping off. I'm guessing you could do both surfaces though.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.