Some photos of my thermal mod of a 2015 retina MacBook

[staceyu]

First time poster here, and I went ahead and did the mod while I was replacing the USB-C port, I have a 2017 i7. I used conductonaut Liquid Metal on the CPU die -> Copper shim, and regular Corsair TM30 thermal paste on the heatsink since it's aluminum, otherwise I would have used Liquid Metal there too. I did NOT use thermal pads on the heatsink so the heatsink does not have any more direct contact with the outer casing vs stock.

At the absolute highest, the temps of the CPU gets to 80-85 C during Cinebench testing, but with less throttling, and it cools rather rapidly. During regular tasks it hovers around 45/50 C, so definitely an improvement from having one mediocre task sending temps to the 70s. Battery temp doesn't exceed 40 C but I have a suspicion that 40 is the highest that the battery thermometer will read, so I can't say the true temp readout is accurate.

Anyway, my Geekbench score: https://browser.geekbench.com/v6/cpu/6646961
Cinebench 2024 score is 48 single core, 88 multi core. I'll do more prolonged testing on Cinebench but I'll happily take the 20% boost in performance just for switching to Liquid Metal and adding a copper shim.

 

[aespana]

I'm thinking about doing this again.

I did it like 2 years before but suffer that problem with the trackpad losing the clicking effect (it worked fine but without the clicking feedback). It repaired itself magically so I don't want to make the process again and loose the clicking feedback forever.

I think is worth it since the Macbook is getting hot sometimes but that feeling or not felling the clicking feedback was insane.

 

[staceyu]

I'm thinking about doing this again.

I did it like 2 years before but suffer that problem with the trackpad losing the clicking effect (it worked fine but without the clicking feedback). It repaired itself magically so I don't want to make the process again and loose the clicking feedback forever.

I think is worth it since the Macbook is getting hot sometimes but that feeling or not felling the clicking feedback was insane.

Likely, the trackpad connecting ribbon is bad. It's a straightforward fix and the part itself is relatively cheap.

As for the mod I used some GPU thermal padding and honestly the outright performance is worse due to thermal throttling, but the CPU itself doesn't really exceed 60 C. One thing that I did notice while testing is that during sustained performance the OS will quickly curb performance and give a temperature warning if it senses that battery temperatures are getting out of hand. I'm not really expecting to get a lot of sustained performance out of the machine (This is the wrong Mac if that's the goal) but Liquid Metal and a copper shim would be your best bet if you're looking for the best performance. It seems that Apple already had in mind that the battery could see some extra wear and tear from the heat of the CPU and built some software safeguards to keep temps in line.

As for what's best for the longevity of the components, I honestly couldn't tell you. But I'm guessing that the 2017 chipsets are more resilient, and you really shouldn't be doing heavy sustained work since that's outside the case for this laptop. For sure the best thing you can do for the rMB is to get rid of that terrible thermal paste and add a copper shim to the aluminum heat spreader. That mod by itself will save 10 C at a minimum and still keep the laptop lap friendly. Even during sustained testing on Cinemark, just changing to Liquid Metal kept temps at 85 C, if not lower.

 

[EIdoradoSZ]

First time poster here, and I went ahead and did the mod while I was replacing the USB-C port, I have a 2017 i7. I used conductonaut Liquid Metal on the CPU die -> Copper shim, and regular Corsair TM30 thermal paste on the heatsink since it's aluminum, otherwise I would have used Liquid Metal there too. I did NOT use thermal pads on the heatsink so the heatsink does not have any more direct contact with the outer casing vs stock.

At the absolute highest, the temps of the CPU gets to 80-85 C during Cinebench testing, but with less throttling, and it cools rather rapidly. During regular tasks it hovers around 45/50 C, so definitely an improvement from having one mediocre task sending temps to the 70s. Battery temp doesn't exceed 40 C but I have a suspicion that 40 is the highest that the battery thermometer will read, so I can't say the true temp readout is accurate.

Anyway, my Geekbench score: https://browser.geekbench.com/v6/cpu/6646961
Cinebench 2024 score is 48 single core, 88 multi core. I'll do more prolonged testing on Cinebench but I'll happily take the 20% boost in performance just for switching to Liquid Metal and adding a copper shim.

My principle is one step at a time, so I'd just repasted thermal grease with Noctua NT-H1, and a 15x15x0.3mm copper in the middle, test. The results are unexpectedly good, peek temperature during Cinebench is around 78 Celsius, and most of the time only around 55-60. Idle temp 35-45. So I think I'm settled here. After all, when the heatsink with thermal pad direct contact with the back, the whole back aluminum case will turn hot a lot, and the SSD and the battery won't like it.

 

[turbo_nine]

After reading various threads about thermal concerns since picking one up for very cheap a few weeks ago, I completed this modification last night on my 2017 i7/16/512 running Sonoma with OCLP. Prior to modification the laptop CPU temp would spike to 100C sort of easily, real-world performance seemed kind of weak and in general the laptop could not quite live up to its promise as a take-anywhere MacBook.

I am pleased to report that the laptop now feels much more usable.


I did not document my process with photos or video, but I can emphasize a few things mentioned already in the thread.

• It's easy to break stuff in here, especially the cables and connectors. Consider your comfort level working on mobile devices and give yourself time without distractions if you aren't a pro.
• I used the plates and pads suggested in the OP for the 2015 model. The CPU die and heat sink are differently shaped and sized for 2017 so as others mentioned some modification to the copper plates and thermal pads are required. I would advise taking special care to re-flatten your partial copper plate after cutting it and stick the cut edge of it on the outside of the 'sandwich' so any flaws in the cut do not create a gap.
• I used MX6 thermal paste which is fairly thick, directly applied to CPU die and to the heat sink. The thick paste retains the copper plates nicely for assembly. Excess compound does allow the plates to shift so take care not to use too much.
• In reassembly, I took care to 'torque' the heat sink screws in a rotating fashion the way you might with automotive gaskets.
• I applied and trimmed the thermal pads to cover all of the heat sink surface and allow case contact but I did not strip away any of the black material on the inside of the bottom/battery case.

Results are better than I expected honestly. The laptop feels noticeably more responsive in extended use. CPU core temps are consistently lower, 40s C at idle and maxing out in the 70s C when running intensive tasks. Battery temps remain at or below 40C (maybe not great) and the 'warm spot' on the bottom of the case is noticeably broader and less hot than before, in other words the bottom case now sinks some heat away from the CPU.

Thanks for writing up and sharing your experiences doing this modification, I love using this thing now.

 

[cyberdogl2]

I was in the middle of performing the steps in this thread. Thanks everyone for your experiences. Especially turbo_nine your post I saw right when I was about to start. I have a similar configuration so that was helpful. A lot of what I'll describe just really echo what turbo_nine experienced.

2017 i7/16/256 running Sonoma with OCLP. Prior to modification it would stay in the range of 80-95C.

• it IS easy to break stuff in there. what i broke was the latch on one of the two connectors leading to the 3.5mm jack. I secured with with kapton tape. i don't use that port so i don't mind if it stopped working one day.
• also used MX6 thermal paste and made sure not to use too much.
• also did not strip the black material inside the bottom case.

One thing I would change if I could do it again is I would get thermal pads that were not precut into squares. I just bought them because I've seen a couple others use them. I thought I'd be able to keep the squares together but they come apart very easily. I would have rather had a solid piece that i could cut exactly in the shape of the macbook's heat sink.

I kept my expectations low, and I'll describe 3 things that I found noteworthy:

1. First the good. I've yet to see this thing go over 70C, which is amazing! Before, light use would have it in the range of 80-95C and sometimes 100C. After a weeks use the difference in temperature is night and day.
2. Second is the surprising: I thought that diverting a massive amount of heat would show up in my thermal scans. The difference isn't as big as I thought. Or, I didn't set a constant range between both pictures. Anyway, I attached a photo of before the modification and after. They look similar, and you could .kind of. see that the heat spread is wider in the 'after' photo. But only by a little bit. Also, the measurement at the brightest spot was about the same temperature. Also both photos were not taken at the same time of day and maybe not the same room temperature. Both were taken with MacBook just booted and idle for 10 minutes. Nevertheless, it's enough to convince me that while internal temperatures have improved greatly, external temperatures spread remains similar.
3. Third is the dissapointing: So I chose this model MacBook exclusively to use with Apple Vision Pro. It's actually thinner than the apple bluetooth keyboards everyone likes to use with AVP. I had hoped that this modification would improve Virtual Display in Apple Vision Pro. It really hasn't.
   
   Based off Intel Power Gadget, I notice that the GPU would sometimes kick on and max out, thus increasing the temperature of the CPU core as well. This made the CPU throttle to 1GHz and the GPU throttle to something like 300Mhz. Which, would grind the Virtual Display to a halt.
   
   My best guess is that Wireless Sidecar uses HVEC encoding (which the Intel HD 615 gpu is capable). But by doing this, my guess is that the gpu is running at 20-40% capacity serving Wireless Sidecar when the MacBook is at idle. Any additional load to the gpu would make the cpu throttle. Further proof is that CPU utilization is about 50% and throttled, so it means that the CPU is not working but is throttled by the gpu heat. Also during this, 80C is the highest reported temperature.
   
   Without Virtual Display, no lag whatsoever everything works fine. Might be a bug in Sonoma loaded by OCLP that doesn't utilize the Intel 615 hardware HVEC encoding to its potential? Not sure, but at this point I don't have the expertise to dig any further. I've tried to disable turboboost and also tried to disable thermal thorttling. Both experiences turned out really bad. Apple really honed the temperature management of this thing. I just wish it was slightly less aggressive.

Like others have said, this is a very very worthwhile modification to do. And not even that hard. Just gotta be extra careful because all the parts and connectors are very tiny.

Thanks to all for sharing your experiences, it made me really confident in attacking this modification