Pro tip: If your MBP is overheating (constant high fan speeds) the problem is usually a lot simpler than the thermal paste issue described in the article. It's usually just dust buildup.
I have fixed multiple overheating Macbook pro's by simply cleaning out the dust buildup underneath the rubber pad, between the fan housing and the heatsink. The airflow compresses the dust into an almost solid piece (like felt), which blocks the airflow. You can usually pull it out as one piece. All you need is a pentalobe screwdriver, a pair of tweezers and maybe some compressed air.
Especially if your laptop often gets into contact with fabrics (on your lap or in bed), you should do this cleaning once a year.
OMG! You were not joking about the "felt" like appearance of the dust!
My MBP seems to always have the fans on. I do occasionally run compressed air across the vents (from the outside) but decided to open it up thanks to this thread.
As I was cleaning, I noticed that when blowing compressed air into the vents, one of the fans was not turning (indicating no air reaching?) I grabbed some precision tweezers and pulled this out: https://i.imgur.com/aIPSRdi.jpg
Nice find, but please don't blow compressed air into a fan without blocking its rotation. It will induce currents and can cause damage to your mainboard.
Even more important: do not blow air in the opposite direction of the one the fan is made to rotate. If you are unlucky you'll damage it, if you are very unlucky it will stop working enterely. Source: I broke a lot of them
Most motherboards (but not all) have basic protection for this, so it's quite unlikely that you break your motherboard electronically that way.
That said, blowing compressed air at a fan will spin the fan much faster than what it was designed for. So you still risk breaking the fan mechanically. Always hold the fan when blowing compressed air at it.
> Pro tip: If your MBP is overheating (constant high fan speeds) the problem is usually a lot simpler than the thermal paste issue described in the article. It's usually just dust buildup.
Another thing also worth looking at is the process list and the Chrome process manager, sorted by CPU. I have no idea what, for example, sueddeutsche.de does - leave a page of them open in the background and the ads will run amok.
Another thing that can cause high CPU load is memory pressure resulting in massive swapping. If 2GB RAM are already set aside for the Docker VM and another 2GB for the core OS, you're left with 12GB RAM and that is not very much.
On the software side of things, I find that Safari treats my laptop a lot better than Chrome. I get maybe 3x better battery life when running with a bunch of active tabs in Safari. Not sure what they're doing to throttle things, but it's pretty effective.
The annoying part is that App Nap also does this with plugged in desktops or even worse with servers that don't even have a monitor plugged in. At least it can be disabled systemwide.
Throttling background apps has its place on a desktop, too - leaving aside power efficiency, there are material performance gains to be made when the OS automatically prioritizes foreground tasks. I wouldn't want FCPX rendering impacted by a rogue banner ad on HowToAddLensFlare.co.cz, know what I mean?
I can see how it might be problematic in a headless environment, but for most people most of the time it's probably a useful feature.
Until some unknown stray process starts to eat 100% of your CPU and you just have no tools to find which tab it is. You can kill the process easily tho, but I'd rather fix problem at it's root.
Outlook consumes 1.5GB RAM alone. Then Mattermost and Slack with each 1.5GB, whoops you're at 7.5 GB. Another 2GB for Photoshop, another GB for Cisco Jabber, Excel... then 50 open Chrome tabs and whoops, where has all the RAM gone?
For people working on 64GB+ workstations this is no problem but Apple only supports 32GB in the newest lineup of MBPs.
My currently open Outlook is holding onto 125 MB of RAM. I have at least 12 spreadsheets open in Excel for 112 MB.
IMHO you are wildly overestimating the RAM use of those programs, aside from the others, as they normally run (sure - I can get Excel over a GB running scripts or manipulating a large file).
2) Screwdriver kit that contains pentalobe and torx bits. You don't need to buy some expensive kit as you can get them cheaply from any hardware store.
Pulling off the bottom plate once you've removed the screws. It's not essential, but it obviates the risk of damaging the case with a spudger or other prying tool.
You can Google that and find many guides on that topic. iFixit has really good guides.
TLDR: Remove the screws on the bottom. You need a special screwdriver and keep in mind that (at least on the <=2015 models afaik) there are different size screws used around the hinge.
ifixit can solve that problem as well (though reasonably expensiveish (additional thing. their magnetic mat with squares on them was the greatest purchase I ever made for the workshop at my first job.))
The ifixit Pro toolkit really is a necessity for anyone even remotely interested in taking things apart.
I've used mine to take apart every conceivable consumer electronic, and through social bits for Nintendo, Apple, etc are priceless.
I also got the magnetic mat, a Christmas gift from my wife, and when disassembling things with lots of screws is useful for drawing out the device for their location.
In case the author shows up here -- I clicked through to the next post (https://bsid.io/posts/2019/03/my-experiment-with-honesty/) which describes what I think is the author getting ripped off and having their work stolen by another individual. If you're reading this, author: you should publicly shame that individual, or outright sue them if you have the resources. In particular you should contact their employer or advisers and let them know that the paper they published is your work, not theirs.
I see Apple hasn't changed at all --- these memorable pages from the official service manual of many years ago shows what they considered the "correct" amount of paste:
The last picture comes from the MacBook Pro 2006 service manual.
I believe the first two come from the MacBook (not pro) early 2006 service manual, since the colour of the paste is white instead of the grey of later manuals which show (only slightly) smaller amounts.
I have some circa 2007-2008 MBP boards lying around that are blue. Based on a quick check of iFixit repair guides, it looks like they stopped using blue when they adopted the unibody chassis.
This is mostly urban legend. You have to really get ridiculous to apply too much thermal paste but you can use too little, so no reason to err in that direction in mass manufacturing. https://www.youtube.com/watch?v=EUWVVTY63hc
> you can use too little, so no reason to err in that direction in mass manufacturing
Exactly. The thermal conductivity of air is 0.026 (W/m/K), your average thermal paste (non-metal) is at 2. For manufacturers it's way better to guarantee uniformly meh conductivity by using too much thermal paste than risking air bubbles by using too little. Most manufacturers will lean this way when they don't just use those POS thermal stickers.
Same with average versus silver: silver provides a gain but a limited one, and it's conductive so you have to make very sure not to leak any onto circuits or you risk a short. That's why manufacturers don't bother.
Is there a meaningful impact of the thickness of excessive TIM? I suspect the answer is no given the clamping power of the mount will get the paste down to roughly a consistent thinness even with over-application.
What about any excess TIM off the edges drawing heat away from the heat pipes/cold plate?
Post author called out the warranty multiple times. If you're in the USA, at least, bear in mind that doing what's described in the post does not void the warranty. The Magnuson-Moss Warranty Act explicitly requires the manufacturer to prove that your action caused the negative impact the warranty claim is made against, otherwise they legally can't void your warranty.
If Apple decides to flagrantly break that law, what options does an individual consumer have? Is there a government agency that will fight on behalf of that consumer, or does the consumer have to hire their own lawyer?
Small claims court can cover up to at least $5,000 or $10,000 in most states, and is generally friendly to the layman or may even ban the use of trained lawyers in some locations.
A representative, usually a regional customer service executive or analogous. They'll often try to deal out first (it may be small claims court's policy to try mediation first).
I contacted MSI about my then new laptop back in 2015. I wanted to add an additional two SSDs (it had an empty m.2 slot and a spinning 2.5" and the OS on a single m.2 SSD).
The problem was that to remove the bottom, I'd have to destroy a warranty void sticker.
I emailed them and they replied in no time at all confirming that removing the sticker to perform the upgrade was not an issue and would not affect my warranty.
I'm in the UK.
It is my understanding that here in Europe, the same rules will be honoured with respect to warranty void stickers so long as you don't cause damage while voiding the sticker to perform whatever it is you're doing.
I figured it was something to that effect, but it's still helpful to call out the consumer protections since a significant portion of Apple's customer base is in the USA.
It's the first thing I do with any new laptop, be it Apple, HP, Dell, or Lenovo. It really makes a massive difference.
In my experience HP pushes the thermal boundary more than most (they generally offer higher TDP options for custom builds than their competitors without tailoring for it).
Some people take this to the next level and decap the processors and then use liquid thermal compound instead of the paste. That probably voids your warranty.
> Some people take this to the next level and decap the processors and then use liquid thermal compound instead of the paste.
They're probably deliding the CPU (removing the integrated heat spreader (IHS) to put the compound on the die directly), not decapping it. I can't imagine a CPU surviving the latter.
The CPU in the article is lid-less so that's not a concern.
And some CPUs have a soldered heatspreader, this allows for much better heat transfer to the IHS but means delidding is destructive (though way less necessary to be fair). Ryzen chips are soldered.
I don't know about those other machines from HP, but the Spectre with i7 standard high-end issue in my office (Big4) is only trouble.
Those are nice if you develop a spreadsheet with some VBA now and then, but once you actually need the power if you're developing enterprise applications you can forget it. Under any serious load they overheat and clock the processor down from 3Ghz+ to 2Ghz.
> This only works efficiently, though, if a very thin layer of thermal paste is applied between CPU and heatsink in such a way that minimises the chance of creating “air bubbles” (air has a bad thermal conductivity).
The issue here isn't really air bubbles. It's true that air has very poor thermal conductivity, but the real issue is that the thermal paste itself has very poor thermal conductivity compared to the solid copper heatsink. The paste is really just supposed to fill in any microscopic pits in the surface to get the copper in good contact. People go to great lengths to make the surfaces as smooth as possible, going as far as to hand lap their heatsinks to a mirror polished finish.
Indeed, TIM is a poor heat conductor compared to metals. If surfaces were ideally polished, TIM would be absolutely not necessary. We’ve done this and saw significant improvement in thermals. Next best thing is liquid metals that have an order of magnitude higher thermal conductivity than anything like arctic silver. Big problem though: they are also electrically conductive so if a drop gets under CPU pins that’s the end of it.
Without going these exocitic routes, the goal of using TIM is to fill up gaps (due to surfaces not being ideal) to replace no thermal conductivity in those places with some thermal conductivity. What you do not want to do, however, is put TIM where there was already good metal-on-metal contact. That’s why too much TIM is definitely bad.
The maximum thickness of the paste layer is determined wholly by the clamping pressure and coplanarity of the thermal interface. You can only end up with an excess of paste between the CPU and heatsink if a) the heatsink contact plate or CPU package aren't flat or b) the clamping mechanism isn't properly pressing them together.
Apply too much thermal paste and it'll just squeeze out of the edges, which is merely inconvenient as long as the paste is reasonably non-conductive. Apply too little paste and you'll get air gaps, which have terrible thermal conductivity. If in doubt, just spread the paste to ensure that it fully covers the contact area.
Yes, you always want the minimum amount which will ensure good contact across the whole surface. You should be pressing (hard) to mate the surfaces and squeeze out excess thermal transfer medium anyway. Desktop/server CPU sockets typically have a mechanical lever action to do that for you and secure the package in place.
Without seeing results AFTER the dust was removed and BEFORE the new thermal paste was applied, it is impossible to attribute the resulting improvement to the thermal paste itself. That computer is super dusty, and I suspect the significant amount of dust removed from the fan and inbetween the fins of the heat sink would explain a large portion of the difference.
If the author has results from after the cleaning, they would help paint a clearer picture.
I guess the dust wasn't that significant to affect the results. This can be seen in the post-Thumbnail. The image was taken immediately after opening the backplate.
I have the same model. Had overheating problems. Took it to the Apple store. They fixed it and refurbished it for me. Even replaced the bottom case because of a couple of dents. It’s like a brand new machine. Charged me $0. This is why I love Apple and will never switch.
You love apple because they don't repair their products, they just trick you into paying for insurance that lets them swap huge chunks of your device instead of finding and fixing what's wrong.
I swapped PTIM (stock thermal paste) with Gallium alloy (Grizzly Conductonaut or Coolaboratories Liquid Metal). The results were absolutely incredible, better than what this article exemplifies. If you're going to open up the chassis, might as well go to the full extent and replace thermal paste with liquid gallium.
I've always wondered why hardware manufacturers don't use better thermal paste on CPUs. When I build PCs I would always replace the stock thermal paste and the cost was always pretty minimal and that was without the economies of scale they'd have. If this small thing helps so much why wouldn't they use it? What is the down side?
What I've heard is that CPU manufacturers (intel specifically in this case) optimize for consistency over time - the crappy intel thermal paste will still be the same level of crappy in 10 years time but a lot of (much) more effective enthusiast stuff might degrade and stop performing as well before then.
Of course I don't remember where I heard this, nor do I have any sources...
The gain is low, it increases costs, and better thermal pastes are electrically conductive so applying them industrially is a more complex endeavour (as you absolutely can't risk leaking/spreading them to any component). The risk/reward just isn't there for manufacturers.
The problem is that they like each other too much. Liquid Gallium basically soaks into the Aluminum, and the Gallium-Aluminum alloy is very weak, at least at room temperature.
It cannot get out unless you overflow it with more than necessary.
Normal shock levels range in 2-5g maximums when you are carrying around your laptop in a briefcase, backback, by hand, etc. This is tested on a shaker table in the reliability lab at 10ms classical half-sine shock. They also do RMS, sine-burst, etc waveforms to ensure that the laptop is going to survive normal use.
The capillary action between gallium liquid sandwiched between the copper heatsink and the die is higher than any of the shocks or external forces you can experience in normal use. Higher by an order of magnitude. Also, the die surface has extremely high surface energy - it wants to wet badly.
It is not going to spill all over the motherboard.
UNLESS - 1) You apply more than necessary. 2) The heatsink is not secured correctly 3) During the application process, you spill all over the motherboard (what Linus @ Linus Tech Tips did in one of his videos).
There is literally no need to worry - I wish I can disclose where I work to give some credibility.
Normal thermal paste is conductive also, including the AS5 that the author uses in the article. You can get electrically nonconductive paste too, but it's not as thermally conductive.
Gallium also destroys aluminum, absorbs into copper (heatsinks), and conducts electricity making it easy to short out components... if I recall correctly.
It absorbs at an incredibly slow rate. If you’re going to apply it, you have to apply an extremely small amount so you’re not going to short anything out.
Most Apple spec intel CPUs have no die-side caps, you’re going to be fine.
A lot of the premium thermal paste products are non-conductive (electrically) and still way better than the stuff they put on at the factory and will improve you situation.
I don’t think that thermal grizzly products are a bad place to start looking. I haven’t tried them myself, but GamersNexus are advertising/recommending them and I do trust them.
you should. kryonaut and conductonaut are both great.
However I'd not recommend liquid metal (gallium) for daily use in a laptop with low tension force. I use liquid metal on delidded ones and kryounat elsewhere.
Is there no chance of that gallium alloy amalgamating with other metals it might encounter? It seems really sketchy, but I guess it's probably not since I've not heard horror stories.
I understand that it doesn't make the system look better to a low-information buyer. So Lenovo aren't doing it. But Apply seem motivated to add small costs in exchange for moderate (if invisible) benefits.
The problem is rarely poor application of paste, but degradation of the paste's performance over time. The paste can chemically degrade, or it can be pumped out of the interface by the movement caused by repeated thermal cycles. Many manufacturers use a solid thermal interface material rather than a paste, which has poorer initial performance but more consistent performance over time.
I would disagree - to give an electronics example, this level of work is done all the time at large scale on PCB manufacturers with solder mask, solder paste, silk screens, etc.
Print shops can do this kind of work, too (precise amounts, sub-millimeter accuracy, irregular surfaces, almost any kind of coating, etc.).
They could, but when you open up a laptop where the manufacturer used thermal paste and you'll more often than not find it apparently applied with a turkey baster in a tenth of a second. Just a big old splat.
I did the same with my 2011 Macbook Pro in 2017, also to similar effect. As another poster points out, the more frequent maintenance task that makes a big difference is removing the fan and pulling the dust out.
With that, my 2011 2.7 GHz i7 Macbook Pro continues to be completely usable even for C++ development (also has a 1 TB SSD and 8 GB of RAM), and hopefully will get me through to a point that Apple releases a laptop that I actually want to buy. If they haven't released a Macbook with a better keyboard by the time this thing dies, it'll be my last Apple laptop.
I'm not sure these numbers are quite right, it looks like the author is just taking the instant values Power Gadget shows even when the clock and utilization are quite different at that particular moment - measured utilization is about a factor of two different. It would have been saner to dump the Power Gadget log and munge it a bit over some time window for more meaningful comparative data.
I should have planned it better and taken more before and after data with fan speeds. I was a bit worried if anything would have gone wrong this experiment would be an expensive mistake. Fortunately, everything was fine in the end. I have been using my laptop since I could see the difference. The Fans don't kick in that often and it had a noticeable impact on battery life.
The effect is so pronounced that a watercooled computer running under full load will typically use less power than an identical aircooled computer, because the lowered silicon current leakage more than makes up for the power consumption of the pump and additional fans.
I would have liked to see more photos of the paste being applied. And is that the CPU and GPU sharing the same heatsink? Or is there another one that needed to be taken apart?
Can I bring my MBP which is under warranty to the Apple Store along with the paste and ask them to do it? I tried bringing my MBP here in EU to an Apple Store for thermal issues and they told me everything it’s fine while it obviously isn’t and they weren’t wiling to do anything but to keep it for “weeks” to see if there’s any issue through a more in depth review.
Very unlikely. If you can convince them of a repair, they’ll want to use their own supplies.
This reminds of Canadian CBC National investigation into how instead of identifying problems, Apple was caught misdiagnosing and pushing purchasing new products.
https://youtu.be/_XneTBhRPYk
I don’t know if you’d have to pay, I suspect you would, but an independent repair shop is likely going to care more about you and your nuanced issue.
They will refuse to do it with 100% certainty. Apple is absolutely obsessed with control, and they go to pretty extreme lengths to prevent people (especially shops) from repairing their devices.
Is it possible to have someone experienced do this for you or your company in San Francisco? Either cleaning out the dust, and/or applying new paste? (how much would it likely cost if so?)
Yes it does — and it’s simple enough that most people can do it, provided that you agree to not hold them responsible if things break. Just go to any tech repair shop and bring them your own thermal paste, which is like $5.
Or better, find a friend to do it for you for a beer. This is just basic maintenance, so very low skill kind of thing.
The reason I ask is because I specifically am interested in those who have the professional confidence to be held responsible if something goes wrong (even in complex situations; e.g. a warped chassis).
Normal Thermal pads are TERRRIBLE and will cause the computer to overheat.
Except those new fancy graphite thermal pads ( thermal grizzly sells some ). Those are generally worse than good paste but are pretty good and they have the added pro of being reusable and keeping their thermal property forever unlike paste. In a laptop that is subject to tension stress they may provide better cooling over time.
undervolting the CPU (with Intel extreme tuning) can also help without opening up the notebook. unfortunately that's Windows only. not sure what is for macos...
eh. instead of buying $400 screwdrivers and opening up the chassis to mess with the thermal paste, i've fixed my 2015 mbp thermal issues by getting a $15 usb laptop cooling pad on amazon.
works like a charm, even with multiple editing sessions on many monitors.
edit: i misread the price of the screwdrivers. I still recommend a cooling pad. High power draw is usually when I'm using extra monitors at a desk anyways.
I had to do a double-take when the he said the paste cost "740". I don't know much about thermal paste so I could've believed it was dollars if it was really something special. But then I noticed the "Rs." beneath the image.
I have fixed multiple overheating Macbook pro's by simply cleaning out the dust buildup underneath the rubber pad, between the fan housing and the heatsink. The airflow compresses the dust into an almost solid piece (like felt), which blocks the airflow. You can usually pull it out as one piece. All you need is a pentalobe screwdriver, a pair of tweezers and maybe some compressed air.
Especially if your laptop often gets into contact with fabrics (on your lap or in bed), you should do this cleaning once a year.