How do I know if my MBP will benefit from heatsink paste reapplication?

Question

Dear Mac Users
I know the theme of hot MacBook Pro's has been done to death, but I would like to add my two pennyworth with respect to the specific question: How do I know if my MBP will benefit from reapplication of the heatsink compound? I have trawled numerous very long threads here and not found a satisfactory answer. The reapplication of the heatsink paste is a bit of a schlep and not without some risk of breaking it, so it would be good to know if the benefit is worth the risk.
One common question is: "Are my MBP temperatures unusual?". A useful mac temperature database can be found at the following link:
http://www.intelmactemp.com/list
Consulting this list should set some parameter space for many users - there does seem to be considerable variability - suggesting highly variable quality of heat sink paste application by Apple. In my particular case I have an early 2011 MBP (i7, quad core, 2.2GHz). This certainly ran very hot. I even had a burn on my thigh from it, which took a while to heal. I know, they are notebooks, not laptops, but still . . . . To put very hot into perspective, the CPU was sitting at over 60C at idle, and whole case would be hot to the touch, such that resting your hands on it was uncomfortable. And yes, the computer was idling, there were no rogue processes chewing up cycles. At full load, the CPU would run at 90C and the whole thing would get very hot. I avoided running major projects or rendering jobs on it to preserve it from heat fatigue - the warranty is now up.
My first attempt at reducing the temperatures, was to clean it. The clue was in the USB/Firewire etc ports. All were very fluffy, which suggested that there was a lot of crud in the heatsink fins. Fixing this was simple; adopt anti-static precautions; remove the back panel; blow air into the exhaust vents at the base of the screen. Numerous large fluff bunnies popped out of the fans and elsewhere. Alternatively, you can unscrew the two fans and partially lift them out (without removing the power cables to the motherboard). Major fluff on the heatsink fins can then be tweezered out, prior to blowing it through, to ensure all debris is removed. I did this subsequently on my Core 2 Duo MBP. After defluffing the idle i7 CPU temperatures dropped from 60C to 43C, but perhaps of more importance was that the whole case was much cooler and the keyboard temperature was close to ambient - much nicer to use.
So that fixed the major heat problem. However, the question remained, would I see any further improvements by stripping the thing down and reapplying the heatsink paste? What I was looking for in this forum was a rule of thumb based on temperature measurements to help me make an informed decision on whether to strip the thing down. As my machine was out of warranty, I wasn't worried about voiding that. Also, where I am there is zero support (Africa), so Genius Bar or similar was not an option. But jiggering a 14month (very expensive) laptop was not something I wanted to do for thrills. The data in many of the posts did not provide any answers. In the end I made some measurements and decided to do the fix. The step by step instructions I followed were very clear and can be found for many Macs/models at:
http://www.ifixit.com/Device/Mac
This guide made life a lot less stressful, as pulling a connector the wrong way could trash the motherboard, and those connectors are so very tiny and fragile. It is very important to take it slow and follow each step to the letter (and not skip steps inadvertently). My before and after heatsink paste fix temperature data are shown below. The idle temps are improved, while the full load temps look like a marginal improvement. However, CPU temperature is not the only story. Certainly after the fix, the whole case is generally cooler, so the heat extraction system is working better. Also, the CPU before and after temps are quite similar at full load, but that assumes the CPU is running at the same power. I do not know enough about the i7 architecture to say if the before speed was being limited by heat, while in the after case the turbo-boost was able to run out to the maximum. Unfortunately, I didn't do any CPU performance tests. So, based on temperatures alone, it looks like a marginal improvement, but it is certainly making a difference to case temperatures.
Note all temperatures were measured at an ambient temperature of 20C on a flat, hard surface using Marcel Bresink's Temperature Monitor. Fan speed was measured using the Fan Control software - set to allow firmware control at idle.
Apple original heat sink paste - big dollop of grey crud and lots of extrusion at the sides.
Idle
CPU 43C
GPU 38C
Heat Sink 2 36C
Heat Sink 3 35C
Fan Speeds 2000rpm
Full Load (all cores maxed with a Boinc Distributed Computing Project (Rosetta - the project, not the Apple technology).
CPU 89C
GPU 57C
Heat Sink 2 52C
Heat Sink 3 49C
Fan Speeds 6200rpm
Arctic Silver heat sink paste
Idle
CPU 38C
GPU 34C
Heat Sink 2 33C
Heat Sink 3 32C
Fan Speeds 2000rpm
Full Load
CPU 86C
GPU 59C
Heat Sink 2 50C
Heat Sink 3 50C
Fan Speeds 6200rpm
Back to my original question: How do I know if my MBP will benefit from reapplication of the heatsink compound? I think temperatures alone will not give a good answer - mainly because the CPU probably has variable output - due to turbo boost. One thing I did notice is that the temperature response of the CPU when it gets switched to full load, does vary according to the quality of the heatsink paste. With the original Apple paste, when I switched the CPU to full load (from idle) the temperature of the CPU went up to its maximum almost instantaneously and stayed at that level ie the graph of temperature vs time was a step function. This suggests that the CPU may be controlling the temperature by throttling itself. Otherwise, as the heatsink warmed up, the CPU should also rise in temperature - mine stayed bang on 90C and fluctuated only by a degree either side.  After application of heat sink paste, the CPU has a much larger effective thermal mass and consequently heats up much more slowly. Turning the CPU up to maximum from idle resulted in the temperature climbing to a maximum over about 3-4s. After peaking at around 92C it dropped back as the fans kicked in, to around 86C.
So perhaps one way of assessing the quality of your heatsink paste is to ramp the CPU to maximum from idle and look at the shape of the temperature profile. A step function suggests a lousy job and benefit may be had from reapplication. A more gently sloping profile, followed by a dip due to the fans kicking in, may suggest you are in good shape. Obviously, this presupposes that your Mac temperatures are not insanely hot to start with (i.e. high 90sC+). In which case, if defluffing doesn't do the job, then new heatsink paste is almost certainly required.
Regards, BB

Answer 1

I'm sorry but this is too funny to pass up.