Interesting, do you have any before and after tmps. My own 2.3 Mid 2012 Retina runs relatively cool, equally less is more in this case...
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Changed Thermal Paste in 15" rMBP
- Thread starter kelebra
- Start date
- Sort by reaction score
Interesting, do you have any before and after tmps. My own 2.3 Mid 2012 Retina runs relatively cool, equally less is more in this case...
Thanks for the post. This is a very interesting topic you bring up here. I have a 15" Early 2013 Retina and I have no issues with temps generally. 50c under idle and 80-85c under load. Until I start getting really hot and worried, then Ill think about changing my paste but for now, I think I'm just fine.
I don't have any proof but, on these summer days, my house is really hot and I often have 28-30ºC temps inside so I was having a hard time keeping the rMBP cool while not doing much. It was actually you that told me to start using UltraFan and set it to something like 65ºC. Truth is, while having only Sparrow, Chrome, Transmission and the other login apps (Dropbox, Fantastical and such), UltraFan kept on constantly ramping up the fans to keep the temps below 65ºC.
I decided to open up the rMBP, clean it, clean all the compound used by Apple and reapply it myself.
Right now, I have 9 tabs open on Chrome and the apps I mentioned above and iStat is showing 53ºC for the CPU and 44ºC for the GPU (HD4000 being used atm). Notice that I'm not using UltraFan anymore so the fans profile is the default one.
Of corse I still get high temperatures while doing intensive operations but I wasn't expecting less than that. It might be possible that the temperatures stall at a lower point when the fans go full throttle.
Thx, I am really not against the idea in principle, equally I don't prescribe that all MBP`s systems have issue, as my 15" MBPr is now out of warrantee I will try it for myself. Right now I am one day from travelling to Borneo for the next four weeks, once I get back I will give it a shot, as temperatures here in Southern China can be similar to yours.
My own MBPr is a 15" Mid 2012 2.3 and it rarely ramps up the fans under moderate load running UltraFan set at 65C, of course this is the concern that all Mac`s are not equal. In some respects I feel that Apple is moving to a very bad place.
Right now my weapon of choice in the field is a Lenovo ThinkPad, Apple portables are simply too fragile to survive the rigours of the field, sadly I am now looking at my next 15" portable Workstation being PC based due to the myriad options and durability, as Apple now only serves the consumer in the portable market
Last edited:
Mine is still under warranty (in the EU you have the right to a 2 year warranty on all products) so I took a risk by doing it but I knew the process and couldn't find any way for Apple to notice that I messed up with the rMBP.
It's not every Mac that is different.. it's every machine. Even same model smartphones will act differently when you overclock them for example. I don't really like that the MBPs can turn to be quite uncomfortable to work with due to heat but I will only start seriously worrying when I see it shutting down due to high temps. *knocks on wood*
And you may consider PCs again.. you'll have several aspects where they win but you'll probably end up buying a MBP again. Personally, I don't see myself going back. At least for now. But if they keep on almost just worrying with iOS, I'll start seriously thinking about it. You're right on that.
It's not every Mac that is different.. it's every machine. Even same model smartphones will act differently when you overclock them for example. I don't really like that the MBPs can turn to be quite uncomfortable to work with due to heat but I will only start seriously worrying when I see it shutting down due to high temps. *knocks on wood*
And you may consider PCs again.. you'll have several aspects where they win but you'll probably end up buying a MBP again. Personally, I don't see myself going back. At least for now. But if they keep on almost just worrying with iOS, I'll start seriously thinking about it. You're right on that.
Agreed all systems are different, marginally by hardware, significantly by software. PC`s are tough one; 7 definitely feels clunker and aged compared to OS X, equally it`s far from unreliable and performs extremely well, 8 I have no experience of. For me a Computer is a tool and it must be fit for purpose, both software & hardware.
I simply prefer OS X, equally I can get Windows to run reliably without issue with a little work. As for OS X the continuous integration of IOS is not welcomed, have already been down the IOS route and discarded it as useless for my needs.
As for Apple`s hardware fit for "Starbucks" I rather hoped Apple would have invested in composite materials, aluminium other than feeling nice is probably one of the worst materials to use for a portable computer that is going to be knocked about....
Applying thermal paste is a wildcard - it can work well, not at all, or make things worse. Nor is there one perfect method to apply the paste. Each heat sink is different in terms of quality, how flat the surface is, etc (if that makes any sense), so more paste may be needed for your particular setup. Also make sure you're lining up the screws to the screw holes on the heatsink exactly (there is some room for error there, so the heatsink doesn't line up exactly, just be careful).
Personally I've repasted my rMBP 3 times. The first time I used a vertical line for the CPU and a grain of rice for the GPU, and let the pressure from the heatsink spread the paste. I saw an overall drop of about 5 degrees C, idle to load. In retrospect I should have been happy with that, but stupid me was unhappy, I wanted better results. So the second time I took the time to tint the heat sink and spread another thin layer with a credit card. My idle temps (rMBP in sig) were high 30s C, and I was happy - until load happened, then I'd temporarily shoot into the 100s playing flash games in Safari, which is completely unacceptable. So the third and current time I went back to the line method and I'm back to where I was initially. I'm idling now in the low 40s, and load high 80s, occasionally hitting low 90s. I see 100s in stress tests, but that's about it.
So I'm done with pasting for now. But ultimately there is no one perfect way of doing it.
Also, just another word of caution (kind of off-topic, but nonetheless). One of the reasons I repasted in the first place is because I thought the CPU was throttling in games due to high temps. Skyrim in particular was a pain in the a** because I'd micro-stutter all over the place. After doing some more reading I've discovered some alternate theories as to why the MacBook throttles. This is a good discussion:
http://forums.anandtech.com/showthread.php?p=35325843
The short of it is if you notice you're dropping frames in games, you might want to try reducing the load on the power supply. For example, once I stopped using my huge external monitor and removed my external HDD and 4 devices from my USB hub, my gaming performance actually improved quite a bit, and my Skyrim problems went away.
Just something to chew on.
Personally I've repasted my rMBP 3 times. The first time I used a vertical line for the CPU and a grain of rice for the GPU, and let the pressure from the heatsink spread the paste. I saw an overall drop of about 5 degrees C, idle to load. In retrospect I should have been happy with that, but stupid me was unhappy, I wanted better results. So the second time I took the time to tint the heat sink and spread another thin layer with a credit card. My idle temps (rMBP in sig) were high 30s C, and I was happy - until load happened, then I'd temporarily shoot into the 100s playing flash games in Safari, which is completely unacceptable. So the third and current time I went back to the line method and I'm back to where I was initially. I'm idling now in the low 40s, and load high 80s, occasionally hitting low 90s. I see 100s in stress tests, but that's about it.
So I'm done with pasting for now. But ultimately there is no one perfect way of doing it.
Also, just another word of caution (kind of off-topic, but nonetheless). One of the reasons I repasted in the first place is because I thought the CPU was throttling in games due to high temps. Skyrim in particular was a pain in the a** because I'd micro-stutter all over the place. After doing some more reading I've discovered some alternate theories as to why the MacBook throttles. This is a good discussion:
http://forums.anandtech.com/showthread.php?p=35325843
The short of it is if you notice you're dropping frames in games, you might want to try reducing the load on the power supply. For example, once I stopped using my huge external monitor and removed my external HDD and 4 devices from my USB hub, my gaming performance actually improved quite a bit, and my Skyrim problems went away.
Just something to chew on.
Did someone try Indigo Xtreme? It uses a different approach and seems to be the best in the TIM market...
If the system is under load a lot, would certain pastes be better than others for regular users? I run BOINC all the time on my mini, which I plan to replace with a rMBP this week.
Pretty cool man, I don't think that I will change the paste until I see significant temp changes that I would need to quickly the heat issue.
But good job on keeping your mac at its best condition.
But good job on keeping your mac at its best condition.
I once saw a pretty good tutorial about how to apply the optimal amount with some trial and error.
Generally you want as little as possible but still enough so there is no dry spot around the center. Just use what you think is a very little press the heat sink on and if it spreads enough, you can probably use less. If it spreads too little (dry parts on the heat sink or chip), use more.
I suppose how much you actually need depends on how flat the two surfaces are. The flatter the less is needed. Metal to metal is what we want for as much as possible.
Generally you want as little as possible but still enough so there is no dry spot around the center. Just use what you think is a very little press the heat sink on and if it spreads enough, you can probably use less. If it spreads too little (dry parts on the heat sink or chip), use more.
I suppose how much you actually need depends on how flat the two surfaces are. The flatter the less is needed. Metal to metal is what we want for as much as possible.
What about this benchmark which is more recent:
----------
Here it is: How to correctly apply thermal paste
I liked the one where they tested mayonnaise as thermal compound.
I have mayonnaise on hand! Alternatively, they also showed that chocolate increased CPU temps. Fun! Now they should test jello or something I liked the one where they tested mayonnaise as thermal compound.
Lol pretty crazy tests they did...
Some of the stuff is probably just seemingly so good because it spreads out better. The lower viscosity of mayo probably means you get initially almost perfect spread because it really does press out every bit of excess mayo. That is probably more of a short term effect.
There are quite a few thermal paste tests on the web that show different rankings and results and a big reason is probably just that application is different. Some are more difficult to apply or just need a slightly different approach. Much more important than the thermal paste is the right application and probably the heat sink surface.
I did mine twice with almost no improvement at first and now a quite big one. I will post some numbers later. Same paste but just doing a better job the second time around.
There are quite a few thermal paste tests on the web that show different rankings and results and a big reason is probably just that application is different. Some are more difficult to apply or just need a slightly different approach. Much more important than the thermal paste is the right application and probably the heat sink surface.
I did mine twice with almost no improvement at first and now a quite big one. I will post some numbers later. Same paste but just doing a better job the second time around.
Just wanted to chime in to say that the "grain of rice" sized thermal paste idiom was coined back in the days when chips were about a square inch in size or more. The amount you put also depends on how smooth the heatsink surface and chip surface are. If the chip is shiny and reflective then it's already very smooth and not as much paste is needed. Adjust accordingly. Most people put too much.
Yes that was probably true for desktop heatsinks but in case of my 2010 MBP there are three chips are 2 are really small. One needs about as small a portion as one can get out of the tube. It is like a half of a quarter of small rice corn.
The chips are complete mirror finish shiny but the heat sink is very poor.
I tried lapping the heat sink which probably was responsible for half the improvement. The heat sink was not very flat at all. The GPU part was falling of to one side. The CPU part (the hottest place) had a recess. If Apple forgot about thermal paste and just gave the heat sink a better finish it might help a lot.
I have a 2010 MBP maybe they did something about that on the retinas or newer MBPs.
I can get 70-73C on the CPu at 4100rpm while playing 1080p video on a second monitor. Previously I had it running at 78-79C at 5998rpm.
Idle with dGPU active went down from 65 to 55 on average.
Overall the notebook is just quite a bit less noisy. It settles at lower rpms, cools down faster and doesn't go as quickly into high rpm.
I have yet to check windows gaming temps but I don't know how comparable the temps are as I have no way of reading out the fans in Windows.
At full load it still settles at the same cpu temps but I did the whole thing mainly to get a notebook that doesn't get noisy just because I attach an external screen and add minor load. I get 3300-3500 where I used to get 4000-4400 rpm fans.
I think fixing the heat sink was most of what matter because otherwise I simply couldn't have used as little paste as I could do now. The recess around the cpu would have been a problem with too little paste. After pressing the heatsink down hard and wriggling it around I removed it and checked how the spread was. On the CPU I managed to press away almost all which means there is really and extremely thin film between that chip and the sink.
The chips are complete mirror finish shiny but the heat sink is very poor.
I tried lapping the heat sink which probably was responsible for half the improvement. The heat sink was not very flat at all. The GPU part was falling of to one side. The CPU part (the hottest place) had a recess. If Apple forgot about thermal paste and just gave the heat sink a better finish it might help a lot.
I have a 2010 MBP maybe they did something about that on the retinas or newer MBPs.
I can get 70-73C on the CPu at 4100rpm while playing 1080p video on a second monitor. Previously I had it running at 78-79C at 5998rpm.
Idle with dGPU active went down from 65 to 55 on average.
Overall the notebook is just quite a bit less noisy. It settles at lower rpms, cools down faster and doesn't go as quickly into high rpm.
I have yet to check windows gaming temps but I don't know how comparable the temps are as I have no way of reading out the fans in Windows.
At full load it still settles at the same cpu temps but I did the whole thing mainly to get a notebook that doesn't get noisy just because I attach an external screen and add minor load. I get 3300-3500 where I used to get 4000-4400 rpm fans.
I think fixing the heat sink was most of what matter because otherwise I simply couldn't have used as little paste as I could do now. The recess around the cpu would have been a problem with too little paste. After pressing the heatsink down hard and wriggling it around I removed it and checked how the spread was. On the CPU I managed to press away almost all which means there is really and extremely thin film between that chip and the sink.
I had a little (2cm by 6cm) wet sharpening stone for knives. I think it is supposed to be 6000/8000 grit (It is a finishing stone). No idea if those grits are comparable to sand paper.
I then used some sand paper on top of glass to make sure the stone is lapped down as flat as possible. Put some the sand paper I had around it and tried to work with the room I had. It takes some care to keep the stone steadily flat on and not tilt it to one side. The heat sink isn't very wide and for the lower grits you need to put on some pressure or you work forever even with copper being fairly soft metal.
Since I only had up to 600 grit sand paper. No idea where to buy finer, I used the stone itself with some water to put on something of a finish. It wasn't a mirror finish but I aimed for flatness first. Wet stones don't work well on the surface with the limited movement.
I would have a sharpening stone here that could probably polish it to a mirror but it is big so I'd have to cut it up and it is also a soft stone which would probably not be best for getting a flat surface as even copper would dig into the stone to much. You are supposed to use those a certain way.
Basically all you need is a flat hard object that fits into the available space and there is enough room for movement to get something done. I used the stone because I could make sure it has a flat surface by simply making it flat.
http://grab.by/qf7Y
This shot is a before picture. Forgot about making an after picture.
The green/yellow area is what you have to work with because the screw holes are higher than the heatsink. The blue is about as big as my stone was.
This is my little stone. With 240 grit sand paper it is flat in about 30 seconds. The black side is really hard the white side softer but finer.
http://grab.by/qf7E
I used black to finish because it would keep its flat from for a long time against copper and doesn't produce as much wet sharpening sand as the white side.
If there is a sand paper on top it doesn't really matter. If the stone is flat to begin with.
I then used some sand paper on top of glass to make sure the stone is lapped down as flat as possible. Put some the sand paper I had around it and tried to work with the room I had. It takes some care to keep the stone steadily flat on and not tilt it to one side. The heat sink isn't very wide and for the lower grits you need to put on some pressure or you work forever even with copper being fairly soft metal.
Since I only had up to 600 grit sand paper. No idea where to buy finer, I used the stone itself with some water to put on something of a finish. It wasn't a mirror finish but I aimed for flatness first. Wet stones don't work well on the surface with the limited movement.
I would have a sharpening stone here that could probably polish it to a mirror but it is big so I'd have to cut it up and it is also a soft stone which would probably not be best for getting a flat surface as even copper would dig into the stone to much. You are supposed to use those a certain way.
Basically all you need is a flat hard object that fits into the available space and there is enough room for movement to get something done. I used the stone because I could make sure it has a flat surface by simply making it flat.
http://grab.by/qf7Y
This shot is a before picture. Forgot about making an after picture.
The green/yellow area is what you have to work with because the screw holes are higher than the heatsink. The blue is about as big as my stone was.
This is my little stone. With 240 grit sand paper it is flat in about 30 seconds. The black side is really hard the white side softer but finer.
http://grab.by/qf7E
I used black to finish because it would keep its flat from for a long time against copper and doesn't produce as much wet sharpening sand as the white side.
If there is a sand paper on top it doesn't really matter. If the stone is flat to begin with.
Last edited:
So I just ordered some 220 grit sand paper. I have the 2013 macbook pro and I think I have wayyyy less space. Do I just rub back n forth as much as I can with water or a circular motion? Goal is to make it flat right?
Last edited:
sand paper doesn't need water but 220 grit is a bit rough. It is fast but you will want to have some 400 and ideally some 800/1000 for finish.
I tried circular but mostly it was back an forth. You just have to work with the space available.
Yes primary goal is to make it flat but it also shouldn't have too rough a finish. I read on some post for heatsink lapping (way easier with normal big desktop heatsinks) that you don't need a mirror finish (1500 grit+) but it 220 helps you make it flat very quickly but it leaves you with quite a rough finish.
I could only buy up to 600 in my local everything around DIY building store (Paint section).
About the water that is only for a certain kind of stones.
Like these.
http://www.fine-tools.com/shapton-stones.html
It is a japanese thing with knive sharpening. The water keeps the metal from heating up and has a few other neat effects but it is only related to how these stones work. Without water they just don't work at all.
I think though that one can by sand paper that is for wet use but not the one I bought. It was just the normal one that you use for would before painting or oiling it.
I tried circular but mostly it was back an forth. You just have to work with the space available.
Yes primary goal is to make it flat but it also shouldn't have too rough a finish. I read on some post for heatsink lapping (way easier with normal big desktop heatsinks) that you don't need a mirror finish (1500 grit+) but it 220 helps you make it flat very quickly but it leaves you with quite a rough finish.
I could only buy up to 600 in my local everything around DIY building store (Paint section).
About the water that is only for a certain kind of stones.
Like these.
http://www.fine-tools.com/shapton-stones.html
It is a japanese thing with knive sharpening. The water keeps the metal from heating up and has a few other neat effects but it is only related to how these stones work. Without water they just don't work at all.
I think though that one can by sand paper that is for wet use but not the one I bought. It was just the normal one that you use for would before painting or oiling it.