Tested: Will the new i7 Mac mini run faster with new thermal paste?

Posted:
in Current Mac Hardware edited December 2018
The new Mac mini is a pint-sized powerhouse, but like other computers, its performance potential isn't reached when it runs too hot. AppleInsider examines if it is possible to improve the temperature situation by re-applying different thermal paste compounds between the processor and heat sink.

Applying Kryonaut thermal paste to a processor
Applying Kryonaut thermal paste to a processor


Apple's 2018 Mac mini is not like the others. It's packing some serious processing power, in a return to the philosophy of the vaunted 2012 model. Plus, there are options for a 3.2GHz i7 processor, 64 GB of RAM and 2 terabytes of incredibly fast storage.






When we tested multi-core performance in Geekbench 4, the Mac Mini scored 25,406 points. That's faster than the 6-core i9 in the 2018 MacBook Pro, and not too far behind the the base iMac Pro.

During testing, we noticed that the i7 Mac mini's CPU reached 100 degrees Celsius shortly after starting our Cinebench R15 5-run stress test. This is basically Intel's hard-limit for temperature, forcing the processor to slow down a bit to cool down and avoid damaging itself.

The Mac mini reaches 100 degrees Celsius
The Mac mini reaches 100 degrees Celsius


When temperatures stabilized, the clock speed stayed between 3.4GHz and 3.5GHz. This isn't bad, but it is lower than we've seen on well-cooled PC systems with the 8th generation 6-core i7 processor.

Based on that observation, it is possible that this processor isn't running at its full potential. Since there's no practical way to replace cooling components like heatsinks with alternatives in the Mac mini, the only real possibility to try is replacing the thermal paste between the processor and the heat sink with a different compound.

Apple with Kryonaut

Warning: This is not a step-by-step guide, and we don't recommend the procedure for most users. Doing so risks damaging the Mac, and will void your warranty.

Before attempting to use a replacement, testing was performed on the default Mac mini setup, giving a base point of comparison. Apple's factory paste scored an average of 1,140 points after five runs of Cinebench R15.

Kryonaut thermal paste from Thermal Grizzly
Kryonaut thermal paste from Thermal Grizzly


The first attempt used Kryonaut from Thermal Grizzly, a highly-rated non-electrically conductive paste that some of the AppleInsider staff has used in PC builds.

Disassembly of the Mini is fairly straightforward but does require specialized tools, in this case items sourced from iFixit. Once we removed the mainboard from the chassis, we had to also remove the EMI shield around the RAM, and the RAM itself to be able to remove the cooler.

  • Removing the mainboard from the Mac mini
  • Cleaning the original factory-applied thermal paste
  • The original thermal paste applied to the Mac mini's processor


The thermal paste that Apple uses was chalkier than we would like to see. We used alcohol wipes to remove the factory paste, then we applied Kryonaut before putting everything back together.

After running Cinebench, the scores were lower than the baseline test. The CPU was now dipping down to 3.3GHz instead of running between 3.4GHz and 3.5GHz, and our five-run average resulted in a lower score, 1,108 compared to 1,140.

Benchmarking the Kryonaut paste directly after application
Benchmarking the Kryonaut paste directly after application


That's definitely not what we were expecting.

We then re-opened the Mac Mini and reapplied Kryonaut using a slightly different application method. After another five runs, we saw an average of 1,108, with the CPU again running at 3.3GHz once thermals stabilized.

Eliminating variables from the previous batch of tests, we switched to the 4K display that was originally attached during our first round of testing for the review, versus the 5K Thunderbolt 3 display we were using. On the lower resolution, we saw a slightly higher average of 1,116 points -- but nothing significantly different.

48 Hours Later

Like many thermal pastes, Thermal Grizzly Kryonaut is labeled as not needing any time to cure. But, with scores like these, we decided to let the paste set over the weekend to see if we could get some better performance.

Benchmarking the Kryonaut paste after 48 hours of curing
Benchmarking the Kryonaut paste after 48 hours of curing


After running all weekend, the average score went up to 1,122 on Monday, the highest we've seen since we replaced the factory paste -- but still lower than what was originally used by Apple.

We opened up the Mac mini again and replaced the paste one last time with Arctic Silver MX-4, another well-rated paste that's half the price of Kryonaut.

Arctic Silver MX-4 thermal compound
Arctic Silver MX-4 thermal compound


Our first run of Cinebench R15's CPU test scored 1195, the highest we've seen so far.

Once the fans ramped up and temperatures stabilized, the CPU stayed between 3.4GHz and 3.5GHz, and at times reached 3.6GHz. Once our 5 runs were complete, Arctic Silver recorded an average of 1,152, a score higher than the original average of 1,140 -- but again, not a lot higher.

Benchmarking the MX4 paste with no curing time
Benchmarking the MX4 paste with no curing time


We then ran Geekbench 4 again, and our multi-core score was 25,600, almost 400 points higher than before.

In Summary

So, all said and done, we do have very slightly better thermal performance using Arctic Silver MX-4 and it could increase a bit once its cured. We'll report back if we see any difference.

We're not sure why the Kryonaut didn't perform well. Most likely, it's possible that it doesn't improve the efficiency of a cooling system in the Mac mini which may already be operating at near maximum capacity.

The paste is designed for a beefier cooling system, and it's possible that the heat dissipation capability that Apple provides doesn't need a higher-end solution and may already be maxed out.

The 2018 Mac mini
The 2018 Mac mini


In Snazzy Labs' testing of the i3 Mac Mini, he was able to get better results swapping the stock paste to another version of Arctic silver, resulting in his Mac mini reaching a maximum of 91C instead of the previous 100C, and running at a consistent 3.6GHz under full load. There are architectural differences between the i3 and i7 -- the i3 doesn't have a "turbo" speed, and the i7 under load will crank out far more heat, and we think that's what the difference is between the two tests.

At this time, we don't think swapping your i7 Mac Mini's thermal paste is worth the risk. Sure, you may get a very little bit more performance, especially with a base model, but at least with the i7 Mac mini, it's not as big of an improvement as is possible on the i3 mini.
«13

Comments

  • Reply 1 of 55
    I think Apple already did the testing of using thermal paste in their labs. Why anyone would do this besides for tests & putting $800 investments into jeopardy is beyond me.
    StrangeDaysrandominternetpersonvulpineSpamSandwich
  • Reply 2 of 55
    noraa1138noraa1138 Posts: 14unconfirmed, member
    I'd love to see a similar test done on the i9 2018 MBP. I really don't have any speed complaints, but when the processor is running full tilt (all 6 cores and 12 threads maxed out), the temp reaches 110° C pretty quickly and the processor throttles slightly to 2.7-2.8GHz (down from the base 2.9GHz). I've thought about reapplying a different thermal paste, but I'm just not sure if it's worth the risk.
    watto_cobra
  • Reply 3 of 55
    Thermal paste aside, maybe additional cooling fans might help? If the paste is getting the heat out to the sink, maybe getting it out of the case is the weakness.
    thttoysandmeradarthekatGeorgeBMacwilliamlondonwatto_cobraSpamSandwich
  • Reply 4 of 55
    DarkStrikeDarkStrike Posts: 1unconfirmed, member
    I think Apple already did the testing of using thermal paste in their labs. Why anyone would do this besides for tests & putting $800 investments into jeopardy is beyond me.
    Actually, one of the easiest best things to do to help thermals is reapplication of the thermal paste. Often times, especially during transit and handling, you may end up with a thermal connection that isn't 100% contacted anymore, and reapplication insures a full high quality contact.
    And contrary to your claim of Apple testing a bunch of different thermal compounds, I would highly doubt they would ever consider putting something as high of quality as kryo or arctic.
    Finally, it's not much of an $800 investment if the peice of crap constantly thermal throttles and is unusable to be used in a prolonged application use, so you might as well try and fix that crap cause you're leaving money on the table allowing it to perform under its rated specs. (Kinda like getting a super fast high end car, but putting a limiter on it that caps at 60mph)
    napoleon_phoneapartcgWerksrepressthishornedfrog
  • Reply 5 of 55
    I often think back to the G5 Pro Towers which if I recall correctly had liquid cooling. I wonder too about phase change options for such, and curious about fan speed/noise with all the new mac hardware, and thoughts on the potential for discrete Vega GPU and possible heat limitations in the new mini...?
  • Reply 6 of 55
    knowitallknowitall Posts: 1,358member
    One percent gain is irrelevant.
    It would help to know the trigger points of the i chip and calculate the max gain you could get before attempting this. I would also use active cooling to max out the result (if it is relevant to do so).
    randominternetpersonwatto_cobra
  • Reply 7 of 55
    I think Apple already did the testing of using thermal paste in their labs. Why anyone would do this besides for tests & putting $800 investments into jeopardy is beyond me.
    Actually, one of the easiest best things to do to help thermals is reapplication of the thermal paste. Often times, especially during transit and handling, you may end up with a thermal connection that isn't 100% contacted anymore, and reapplication insures a full high quality contact.
    And contrary to your claim of Apple testing a bunch of different thermal compounds, I would highly doubt they would ever consider putting something as high of quality as kryo or arctic.
    Finally, it's not much of an $800 investment if the peice of crap constantly thermal throttles and is unusable to be used in a prolonged application use, so you might as well try and fix that crap cause you're leaving money on the table allowing it to perform under its rated specs. (Kinda like getting a super fast high end car, but putting a limiter on it that caps at 60mph)
    As long as the high end car lasts ,I don’t care.
  • Reply 8 of 55
    thttht Posts: 3,230member
    I think Apple already did the testing of using thermal paste in their labs. Why anyone would do this besides for tests & putting $800 investments into jeopardy is beyond me.
    Actually, one of the easiest best things to do to help thermals is reapplication of the thermal paste. Often times, especially during transit and handling, you may end up with a thermal connection that isn't 100% contacted anymore, and reapplication insures a full high quality contact.
    And contrary to your claim of Apple testing a bunch of different thermal compounds, I would highly doubt they would ever consider putting something as high of quality as kryo or arctic.
    Finally, it's not much of an $800 investment if the peice of crap constantly thermal throttles and is unusable to be used in a prolonged application use, so you might as well try and fix that crap cause you're leaving money on the table allowing it to perform under its rated specs. (Kinda like getting a super fast high end car, but putting a limiter on it that caps at 60mph)

    The i7-8700B is running as specified with 3.2 GHz base and 4.6 GHz turbo clocks. It’s not throttling. It’s running as designed, and if the article is right, this CPU was running at 3.4 to 3.5 GHz after it reached peak temperatures. That’s 5 to 10% better than advertised.
    cgWerkswatto_cobra
  • Reply 9 of 55
    thttht Posts: 3,230member

    DAalseth said:
    Thermal paste aside, maybe additional cooling fans might help? If the paste is getting the heat out to the sink, maybe getting it out of the case is the weakness.
    Yup. More fans probably won’t work with the Mac mini, but if they could modify the firmware or modify how the fan runs, they can increase the RPMs on the fans something like 20 to 30% more than baseline and maybe gain some additional performance.

    They probably should run a few tests in a refrigerator or freezer first to see what the potential is. Putting it in a freezer is not the safest thing to do either, but if they are comfortable applying and reapplying thermal paste, sure. Also, paint both the inside and outside of the case black, and don’t leave it anywhere sunny.
    watto_cobra
  • Reply 10 of 55
    hodarhodar Posts: 271member
    At this time, we don't think swapping your i7 Mac Mini's thermal paste is worth the risk. Love the honest reporting. Yes, the result is not what you hoped - you ran the experiment twice, and confirmed everything. This is solid information, and shows that Apple did their homework.
    StrangeDaystoysandmerandominternetpersoncgWerksrepressthisDeelronwatto_cobra
  • Reply 11 of 55
    I think Apple already did the testing of using thermal paste in their labs. Why anyone would do this besides for tests & putting $800 investments into jeopardy is beyond me.
    Actually, one of the easiest best things to do to help thermals is reapplication of the thermal paste. Often times, especially during transit and handling, you may end up with a thermal connection that isn't 100% contacted anymore, and reapplication insures a full high quality contact.
    And contrary to your claim of Apple testing a bunch of different thermal compounds, I would highly doubt they would ever consider putting something as high of quality as kryo or arctic.
    Finally, it's not much of an $800 investment if the peice of crap constantly thermal throttles and is unusable to be used in a prolonged application use, so you might as well try and fix that crap cause you're leaving money on the table allowing it to perform under its rated specs. (Kinda like getting a super fast high end car, but putting a limiter on it that caps at 60mph)
    If you think Apple, whose materials science and electrical engineering is legend, hasn’t consider a quality thermal paste, you’re high. 

    Also, it’s not a piece of crap. It’s a small form factor for particular use cases. If your use cases differ that’s fine, but it doesn’t mean this machine is crap. 
    edited December 2018 mwhiterandominternetpersonDeelronwatto_cobraSpamSandwich
  • Reply 12 of 55
    cpsrocpsro Posts: 2,470member
    Too bad the 2018 MBP is such a nightmare to disassemble/reassemble (iFixit score of 1/10), because AI's CPU-only benchmark of a new VEGA20 model suggested Apple is using a different thermal paste--it scored a few percent higher than the 560X model.
    williamlondon
  • Reply 13 of 55
    thttht Posts: 3,230member
    cpsro said:
    Too bad the 2018 MBP is such a nightmare to disassemble/reassemble (iFixit score of 1/10), because AI's CPU-only benchmark of a new VEGA20 model suggested Apple is using a different thermal paste--it scored a few percent higher than the 560X model.
    No. It does not suggest that. There are many many reasons. I could name several that have nothing to do with thermal paste.
    edited December 2018 williamlondonwatto_cobra
  • Reply 14 of 55
    In my PC overclock days, I used Goof Off to clean the heatsink and the processor to clean the existing thermal paste. Alcohol pads were never enough to clean or make a difference. Also back in the days, I used solid copper heatsinks for better thermal dissipation, not just reuse the same heatsink. Lastly, Artic-Silver at the time requires several heat/cool cycles to change its crystallization structure for best results. I got best results after the first week of normal use.
    radarthekatcaladanianwatto_cobraSpamSandwich
  • Reply 15 of 55
    What about sitting it on top of a bag of frozen peas? ;-) Or, productized, some kit that replaces the base and pushes cold air up into the system? I know I know, I'm pretty sure the cost/benefit analysis of any such "cooling base" product would never work out.
    watto_cobra
  • Reply 16 of 55
    I think Apple already did the testing of using thermal paste in their labs. Why anyone would do this besides for tests & putting $800 investments into jeopardy is beyond me.
    Actually, one of the easiest best things to do to help thermals is reapplication of the thermal paste. Often times, especially during transit and handling, you may end up with a thermal connection that isn't 100% contacted anymore, and reapplication insures a full high quality contact.
    And contrary to your claim of Apple testing a bunch of different thermal compounds, I would highly doubt they would ever consider putting something as high of quality as kryo or arctic.
    Finally, it's not much of an $800 investment if the peice of crap constantly thermal throttles and is unusable to be used in a prolonged application use, so you might as well try and fix that crap cause you're leaving money on the table allowing it to perform under its rated specs. (Kinda like getting a super fast high end car, but putting a limiter on it that caps at 60mph)
    And yet Apple's out of the box solution was better than the "high quality" stuff.  Funny how that works.
    watto_cobra
  • Reply 17 of 55
    Very interesting.  Glad you guys tried this and not me so thanks for that.  Personally I think that Apple probably has done its homework on the best cooling solution but having said that it probably is a compromise that works for most customers as well as improves reliability and reduces warranty repairs.   

    On all of my Apple machines I use fan control software to control my fan speeds.   This works tremendously well and keeps my temperatures down with really very minimal increase in fan noise.  Plus when not needed I can just let it go back to the Apple defaults.  Not sure if this will work on the new hardware for the mac mini but might be something you all can test.  

    The only downside to using the fan control software that I can think of is that it might reduce the lifetime of the fans but I have machines that go back to 2012 and their fans are still working fine.
    cgWerkswatto_cobra
  • Reply 18 of 55
    tht said:

    DAalseth said:
    Thermal paste aside, maybe additional cooling fans might help? If the paste is getting the heat out to the sink, maybe getting it out of the case is the weakness.
    Yup. More fans probably won’t work with the Mac mini, but if they could modify the firmware or modify how the fan runs, they can increase the RPMs on the fans something like 20 to 30% more than baseline and maybe gain some additional performance.

    They probably should run a few tests in a refrigerator or freezer first to see what the potential is. Putting it in a freezer is not the safest thing to do either, but if they are comfortable applying and reapplying thermal paste, sure. Also, paint both the inside and outside of the case black, and don’t leave it anywhere sunny.
    What?  Why would painting anything black help?  There's no light inside the case, so how could it make a difference?
    watto_cobraSpamSandwich
  • Reply 19 of 55
    sflocalsflocal Posts: 4,652member
    I think Apple already did the testing of using thermal paste in their labs. Why anyone would do this besides for tests & putting $800 investments into jeopardy is beyond me.
    Actually, one of the easiest best things to do to help thermals is reapplication of the thermal paste. Often times, especially during transit and handling, you may end up with a thermal connection that isn't 100% contacted anymore, and reapplication insures a full high quality contact.
    And contrary to your claim of Apple testing a bunch of different thermal compounds, I would highly doubt they would ever consider putting something as high of quality as kryo or arctic.
    Finally, it's not much of an $800 investment if the peice of crap constantly thermal throttles and is unusable to be used in a prolonged application use, so you might as well try and fix that crap cause you're leaving money on the table allowing it to perform under its rated specs. (Kinda like getting a super fast high end car, but putting a limiter on it that caps at 60mph)
    You must be using some type of "Alternative Math".  Rounding the cooling improvement of this test to 1% (and that's generous), assuming your "super fast high-end car" is the Bugatti Veyron that can reach 267mph, that one percent equates to 2.67mph loss on the top end.  Not even worth the effort really if it means damaging or voiding the warranty.

    So I would agree with the prior poster.  At that limit, I wouldn't care in the least.

    FYI, I bought the new Mac Mini a couple weeks ago and have been using it every day with dual Thunderbolt monitors.  The case is barely warm to the touch and if the fans turn on, I sure as heck can't hear them.

    60mph cap?  not even close.
    edited December 2018 randominternetpersonDeelroncaladanianwatto_cobraSpamSandwich
  • Reply 20 of 55
    What about sitting it on top of a bag of frozen peas? ;-) Or, productized, some kit that replaces the base and pushes cold air up into the system? I know I know, I'm pretty sure the cost/benefit analysis of any such "cooling base" product would never work out.
    In the oldish days of my PowerBook G4 (not the old days of my WallStreet G3, or *really* old days of my PB170), heat was a real problem. I found a fairly light (1-2 lbs?) metal board in my kitchen, ostensibly to be used to defrost meat faster than just leaving it in the sink. It worked great as a heat sink and heat spreader, with the PB sitting right on top of it. Made it OK to use in my lap (it could really be uncomfortable directly against skin or light clothes). If I didn't want to deal with a loud fan, and I knew I was going to be doing something heavy-duty, I could freeze it first. It would help keep the Mac cool for an hour or two. Something like that might work today too.

    BTW, I have complained a couple of times about badly-written or underinformative articles here, so it's only fair to compliment this one as it is quite the opposite.
    cgWerkswatto_cobra
Sign In or Register to comment.