Mac mini 2018 - thermal solutions thread  

Page 2 / 2 Prev
  RSS

(@anethema)
Active Member
Joined: 1 month ago
Posts: 10
November 13, 2018 4:37 am  

That is the idea though. Right now it is only able to maintain base clocks not boost clocks. The goal isn't neccesarily to keep the CPU staying cool, but to be able to maintain higher clocks in heavy load situations for better performance.

Soon I will add my system & eGPU details or a build link to this my signature to give context to my posts


ReplyQuote
bluetech
(@bluetech)
Active Member
Joined: 1 month ago
Posts: 14
November 13, 2018 6:19 am  

CPU-Throttleing at i7-8700B   is only 4,7%
between idleing or normal useing before bench. 

Cooling: Less than 10 degrees is not allowed 😉

Soon I will add my system & eGPU details or a build link to this my signature to give context to my posts


ReplyQuote
nertopia
(@nertopia)
New Member
Joined: 4 weeks ago
Posts: 1
November 18, 2018 5:56 am  

I built my own semi-custom enclosure for the 2018 Mac Mini. I have two small fans with a manual speed controller and I turn it on when the I hit higher CPU use. Together with Apple's air cooling it runs at low temps, so I am pleased with this setup.

Next step is the eGPU... so many choices so little time.

Soon I will add my system & eGPU details or a build link to this my signature to give context to my posts


ReplyQuote
James
(@jdstephens)
Active Member
Joined: 11 months ago
Posts: 13
November 19, 2018 12:53 pm  
Posted by: nertopia

I built my own semi-custom enclosure for the 2018 Mac Mini. I have two small fans with a manual speed controller and I turn it on when the I hit higher CPU use. Together with Apple's air cooling it runs at low temps, so I am pleased with this setup.

Next step is the eGPU... so many choices so little time.

Are you able to show a picture of the intel power gadget with some benchmarks. Would like to see the same tests with and without your custom setup running.
Of course a photo of what you’ve done would be much appreciated

Soon I will add my system & eGPU details or a build link to this my signature to give context to my posts


bluetech liked
ReplyQuote
bluetech
(@bluetech)
Active Member
Joined: 1 month ago
Posts: 14
November 19, 2018 1:43 pm  

The i7-8700B on the Mac mini 2018 runs at 95,4% ✔️

I've done this to test the cpu:
-4 hours of prime95, 12Threads, 8k for max heat
-than Cinebench 15 -> 1211 pts    result "cold"->1269 pts

Soon I will add my system & eGPU details or a build link to this my signature to give context to my posts


ReplyQuote
James
(@jdstephens)
Active Member
Joined: 11 months ago
Posts: 13
November 20, 2018 2:40 am  
Posted by: bluetech

The i7-8700B on the Mac mini 2018 runs at 95,4% ✔️

I've done this to test the cpu:
-4 hours of prime95, 12Threads, 8k for max heat
-than Cinebench 15 -> 1211 pts    result "cold"->1269 pts

So you saying it reduces speed by 4.6% when its at max temp?

Soon I will add my system & eGPU details or a build link to this my signature to give context to my posts


ReplyQuote
Baupfhor
(@baupfhor)
New Member
Joined: 2 weeks ago
Posts: 4
December 1, 2018 3:28 am  

Im going to liquid metal mine as soon as my ram comes in the mail.

2018 Mac Mini i7 16gb 256gb 10gbe - Akitio Node w/ FSP Dagger 600w psu and Sapphire RX Vega 64
i3 8350k Hackintosh w/ GTX 1080ti


ReplyQuote
ekkim
(@ekkim)
New Member
Joined: 2 weeks ago
Posts: 1
December 1, 2018 4:45 am  

Hey all, I'm a new member here.

We've just bought a 2018 mac mini with a 3.2GHz 6-core i7, 32GB ram, 1TB SSD and Sonnet 650w egpu with a Vega 64. We're curious if we can meaningfully get any more performance from this mac mini before the new mac pro is released next year sometime. The computer is solely used for video editing work in Adobe and FCX.

Is there anything to be gained by reapplying thermal past or perhaps using Innovation Cooling's Graphite Thermal Paste?

We're not really expecting too much more but if there was something worthwhile we would get excited 🙂

Thank you 🙂

Soon I will add my system & eGPU details or a build link to this my signature to give context to my posts


ReplyQuote
mac_editor
(@mac_editor)
Noble Member Moderator
Joined: 2 years ago
Posts: 1586
December 1, 2018 6:21 am  

Seems like thermal paste change helps a lot on the Mac mini.

purge-wrangler.shpurge-nvda.shset-eGPU.sh
----
Troubleshooting eGPUs on macOS
Command Line Swiss Knife
----
Multiple Build Guides


hornedfrog liked
ReplyQuote
Eightarmedpet
(@eightarmedpet)
Noble Member
Joined: 2 years ago
Posts: 1101
December 1, 2018 7:54 am  
Posted by: mac_editor

Seems like thermal paste change helps a lot on the Mac mini.

God I hate youtubers... so overly dramatic and such forced opinions, all done to drive engagement in the chase for the advertisers dollar. 
Maybe I should start a channel, “honest reviews from a non screechy anti drama perspective “. I doubt anyone would watch.

2017 13" MacBook Pro Touchbar + [email protected] (AKiTiO Thunder3) + Win10
2017 13" MacBook Pro Touch Bar + [email protected] (Sonnet Breakaway Box) + Win10

2017 13" MacBook Pro Touch Bar + GTX1070 (debating 2070 upgrade) @32Gbps-TB3 (Razer Core V1) + Win10
---
LG 5K Ultrafine flickering issue fix


hornedfrog liked
ReplyQuote
jim_survak
(@jim_survak)
Eminent Member
Joined: 2 years ago
Posts: 40
December 1, 2018 11:34 am  

I agree with you about the YouTubers, @Eightarmedpet  "It throttled 200MHz down! Oh no there goes all my productivity!"

That being said, I do a little YT work as well. I made a video with my thoughts on it but to spare you I'll give the short version: the new Mac mini isn't meant for the "budget" crowd; it's really no longer the introductory Mac. Something that people like snazzylabs fail to grasp is that the Mac mini just wasn't selling well in the customer space but was doing very well in the enterprise, so Apple re-positioned it for that purpose. As someone who works in an enterprise environment I can tell you that we have no problem dropping $1500 per workstation & ordering 50 or 60 workstations at a time. No major enterprise will have an issue with an $800 workstation.

Sorry for the detour, back on topic: I know it's been discussed at length in other places but my 2012 Mac mini didn't have great thermal solution either. I used Noctua's NT-H1 thermal grease & dropped temperatures down from ~77C to ~65C. Bumping the CPU fan with TG Pro (which someone else mentioned here) up to 2200 RPM bring the temps down to 49C. If others are willing to try different thermal grease options we might could build a running thread of what temperature changes are.

2012 Mac mini: Core i7, 16GB RAM, Toshiba 1TB SSD, Seagate 1TB HDD, Sonnet B350 w/ XFX Radeon RX-480 8GB
Custom: Ryzen 7 1700, 16GB Corasir DDR4-3200MHz RAM, 2x Corsair 500GB Neutron SSD, Seagate 3TB SSD, EVGA Nvidia 980Ti
Alienware 13: Core i5-4210U, 16GB RAM, PNY XLR8 500GB SSD, Nvidia 860m 2GB, w/ AGA


ReplyQuote
Eightarmedpet
(@eightarmedpet)
Noble Member
Joined: 2 years ago
Posts: 1101

jim_survak liked
ReplyQuote
jim_survak
(@jim_survak)
Eminent Member
Joined: 2 years ago
Posts: 40
December 1, 2018 12:21 pm  


I start discussing the Mac mini @ 39:50. It's about 20 minutes because I love the Mac mini & have a lot to say. And I think it needed explanation as to what the Mac mini was & now what it is.

2012 Mac mini: Core i7, 16GB RAM, Toshiba 1TB SSD, Seagate 1TB HDD, Sonnet B350 w/ XFX Radeon RX-480 8GB
Custom: Ryzen 7 1700, 16GB Corasir DDR4-3200MHz RAM, 2x Corsair 500GB Neutron SSD, Seagate 3TB SSD, EVGA Nvidia 980Ti
Alienware 13: Core i5-4210U, 16GB RAM, PNY XLR8 500GB SSD, Nvidia 860m 2GB, w/ AGA


ReplyQuote
Baupfhor
(@baupfhor)
New Member
Joined: 2 weeks ago
Posts: 4
December 5, 2018 4:36 am  

I went ahead and added Thermal Grizzly liquid metal to my i7. I ran both CinebenchR15 and Blender (BMW) and screen recorded before and after. I used Intel Power Gadget to monitor temps and clock speed.  Peak temps were a few degrees cooler but nothing to OMG about, however, it did allow it to turbo boost a couple hundred MHz higher for longer resulting in a higher cinebench score and faster blender render times. I also saw a 200 point increase in Geekbench 3 breaking 5000/27000. All runs were with Mac fan control set to auto 50c fan ramp up. 
Cinebench before 1219cb @3.6ghz 97-100c
Cinebench after 1255 @ 3.8ghz 90-97c
Blender before 5:50 95-97c @ 3.5-3.6 ghz
Blender after 5:23 95-97c @ 3.8-3.9 ghz
Geekbench 3 before 4880/25497
Geekbench 3 after 5066/27383
Geekbench 4 before 5501/22279
Geekbench 4 after 5809/26632

2018 Mac Mini i7 16gb 256gb 10gbe - Akitio Node w/ FSP Dagger 600w psu and Sapphire RX Vega 64
i3 8350k Hackintosh w/ GTX 1080ti


theitsage and hornedfrog liked
ReplyQuote
hornedfrog
(@hornedfrog)
Active Member
Joined: 1 week ago
Posts: 5
December 5, 2018 2:00 pm  
Posted by: Baupfhor

I went ahead and added Thermal Grizzly liquid metal to my i7. I ran both CinebenchR15 and Blender (BMW) and screen recorded before and after. I used Intel Power Gadget to monitor temps and clock speed.  Peak temps were a few degrees cooler but nothing to OMG about, however, it did allow it to turbo boost a couple hundred MHz higher for longer resulting in a higher cinebench score and faster blender render times. I also saw a 200 point increase in Geekbench 3 breaking 5000/27000. All runs were with Mac fan control set to auto 50c fan ramp up. 
Cinebench before 1219cb @3.6ghz 97-100c
Cinebench after 1255 @ 3.8ghz 90-97c
Blender before 5:50 95-97c @ 3.5-3.6 ghz
Blender after 5:23 95-97c @ 3.8-3.9 ghz
Geekbench 3 before 4880/25497
Geekbench 3 after 5066/27383
Geekbench 4 before 5501/22279
Geekbench 4 after 5809/26632

Thank you!  Was this a difficult project?   What are the average temperatures Now?

2018 Mac mini i7 1TB 8G ram.
To add: Sonnet Breakaway Box 550W. MSI RX 570 8G (want to upgrade to Sapphire Vega 64 - 21275-03-20G card)
32G RAM


ReplyQuote
Baupfhor
(@baupfhor)
New Member
Joined: 2 weeks ago
Posts: 4
December 7, 2018 7:21 pm  
Posted by: hornedfrog
Posted by: Baupfhor

I went ahead and added Thermal Grizzly liquid metal to my i7. I ran both CinebenchR15 and Blender (BMW) and screen recorded before and after. I used Intel Power Gadget to monitor temps and clock speed.  Peak temps were a few degrees cooler but nothing to OMG about, however, it did allow it to turbo boost a couple hundred MHz higher for longer resulting in a higher cinebench score and faster blender render times. I also saw a 200 point increase in Geekbench 3 breaking 5000/27000. All runs were with Mac fan control set to auto 50c fan ramp up. 
Cinebench before 1219cb @3.6ghz 97-100c
Cinebench after 1255 @ 3.8ghz 90-97c
Blender before 5:50 95-97c @ 3.5-3.6 ghz
Blender after 5:23 95-97c @ 3.8-3.9 ghz
Geekbench 3 before 4880/25497
Geekbench 3 after 5066/27383
Geekbench 4 before 5501/22279
Geekbench 4 after 5809/26632

Thank you!  Was this a difficult project?   What are the average temperatures Now?

Not too difficult but I’m used to tearing things apart. Average temps are still 90-97c under full load, just slightly higher boost clock speed. Idle temp still 40-41c with fan speed at 1700. The benchmarks show a noticeable performance gain, but Apple still
allows it to run up to 100c.

2018 Mac Mini i7 16gb 256gb 10gbe - Akitio Node w/ FSP Dagger 600w psu and Sapphire RX Vega 64
i3 8350k Hackintosh w/ GTX 1080ti


hornedfrog liked
ReplyQuote
hornedfrog
(@hornedfrog)
Active Member
Joined: 1 week ago
Posts: 5
December 7, 2018 11:16 pm  
Posted by: Baupfhor
Posted by: hornedfrog
Posted by: Baupfhor

I went ahead and added Thermal Grizzly liquid metal to my i7. I ran both CinebenchR15 and Blender (BMW) and screen recorded before and after. I used Intel Power Gadget to monitor temps and clock speed.  Peak temps were a few degrees cooler but nothing to OMG about, however, it did allow it to turbo boost a couple hundred MHz higher for longer resulting in a higher cinebench score and faster blender render times. I also saw a 200 point increase in Geekbench 3 breaking 5000/27000. All runs were with Mac fan control set to auto 50c fan ramp up. 
Cinebench before 1219cb @3.6ghz 97-100c
Cinebench after 1255 @ 3.8ghz 90-97c
Blender before 5:50 95-97c @ 3.5-3.6 ghz
Blender after 5:23 95-97c @ 3.8-3.9 ghz
Geekbench 3 before 4880/25497
Geekbench 3 after 5066/27383
Geekbench 4 before 5501/22279
Geekbench 4 after 5809/26632

Thank you!  Was this a difficult project?   What are the average temperatures Now?

Not too difficult but I’m used to tearing things apart. Average temps are still 90-97c under full load, just slightly higher boost clock speed. Idle temp still 40-41c with fan speed at 1700. The benchmarks show a noticeable performance gain, but Apple still
allows it to run up to 100c.

Thank you,   I will give it a go.  I have some Prolimatech PK3.   It is probably not as strong as the liquid metal but must be better than the stock compound.

2018 Mac mini i7 1TB 8G ram.
To add: Sonnet Breakaway Box 550W. MSI RX 570 8G (want to upgrade to Sapphire Vega 64 - 21275-03-20G card)
32G RAM


ReplyQuote
(@megaseppl)
New Member
Joined: 5 days ago
Posts: 2
December 10, 2018 5:09 pm  

Hi there,


I have had 3 temperature problems with my Mac mini:



1. High temperature in Sleep mode

2. High temperature in idle / normal work

3. High temperature in heavy load.



1.For the first problem I found two reasons: My sleep mode was not working correctly. Because of the missing pulsing LED I was not realizing that the Mac was still running even though external Disks did not spin, Monitors went off etc. But the Mac was still consuming almost 20 Watts, but without a spinning fan. That result in a case top temperature of more than 46°C. More than on heavy load where the fan is blowing most of the heat away.



After 2 hours of measuring the energy consumption I found the reasons:

- The USB-C hub did not let the Mac sleep.

- Same for Power Nap.


Only if I removed the hub AND disabled Power Nap the temps got normal.



2. While normal work (no CPU heavy apps like Browsing, Twitch stream etc.) the Mac mini got really warm. Up to 70° for the CPU. Components like WiFi or SSD were quite hot too. As a solution I built the Glory Blow Hole: A hole inside the table with a 14cm 5V USB fan, controlled by PWM.



If you read Glory Blow Hole, please do it with Morgan Freemans voice in your mind... but with German accent!



 


On a wooden panel with a hole I've placed a 14cm dust filter, then a Noctua 5V PWM fan, controlled by a Noctua PWM controller. All USB powered. The PWM controller will be replaced by a smarter solution wich will be controlled by my home automation later. At presence it goes on/off automatically when the Mac is in use. In future the speed will get adjusted dependent on the temperature of the CPU.


 Above the fan another panel which will later be glued on the table.


This is the final view. Almost not visible...





... except if you lift Kenobi, my Mac Mini. I have added door isolation foam where the Mac is floating on.





View from below. You can see the Mac mini without its plastic cap.



Results: 
The most dramatic improvements can be seen in short. high workloads. Cinebench got an improvement of up to 11 percetanges:




Before: 1130, after 1259.


On long workloads the improvement is much less: At a ~50 minutes test render in Cinema 4D it saved about 2 minutes. But hey, it's something!

Components like WiFi or SSD are way cooler than before.



I have no Geekbench results from before my Glory Blow Hole was implemented, but I did not expect higher than normal results - because Geekbench does not produce much heat. So my second fan is not really helping here.
The Multicore results where as expected in normal range. But the Single Core results seem to be the third highes results ever recorded with a Mac mini in Geekbench (and second highest device):

https://browser.geekbench.com/v4/cpu/11172775 


And because the Mac mini in general has the fastest Single Core performance of all current Macs, it seems that my Mac Mini ist the second fastest Mac on the planet right now! ;-)





3. 
I wanted to improve the heavy load temperatures further (beside my Glory Blow Hole that seemed already to help here) by replacing the Thermal compound. So I cleaned the CPU and heatsink and put Artic MX-2 on it. But honestly that was a waste of time: It did not improve the temperatures or the clock speed at all. Maybe I'll have to wait a little...sometimes it needs time till thermal paste works optimal.



Total work for the Glory Blow Hole took about 3 hours including the time to find the right fan (silent, USB, PWM) and its controller.

So, as a summary I think giving the Mac Mini cool air is more important than changing the thermal compound. This might be different on other Mac minis of course. But it should not be the first thing you try if you want to improve performance. 

Maybe this is a inspiration for some of you. 🙂

No eGPU yet... planning to buy a Razer Core X with ASUS Vega 64 for my Mac mini


ReplyQuote
theitsage
(@itsage)
Famed Member Admin
Joined: 2 years ago
Posts: 3220
December 10, 2018 5:21 pm  

@megaseppl This is very cool! 2018 Mac mini owners with the i7 should pay a visit to your Glory Blow Hole.  😀

Best ultrabooks for eGPU use

eGPU enclosure buying guide

86 external GPU build guides


Juan C. Yunis, Shirokiya, hornedfrog and 1 people liked
ReplyQuote
(@pierre-emmanuel_pardon)
Active Member
Joined: 12 months ago
Posts: 13
December 10, 2018 11:04 pm  

Hi,

New setup :

Working like a charm !

Old hardware: MacBookPro11,5 (15 Retina, i7 4980HQ) dGPU R9 M370X + TB2/TB3 adapter + Razer Core v2 + GTX 1070 Ti

New hardware: Mac mini (2018, i7 8700B, 32 GB RAM, 1 TB SSD)+ Razer Core v2 eGPU + SAPPHIRE Radeon RX Vega 64 + Samsung C27HG70 display
New software: macOS 10.14.2 (18C54), Windows 10 to be tried !


ReplyQuote
Juan C. Yunis
(@juan_c_yunis)
Estimable Member
Joined: 1 year ago
Posts: 116
December 10, 2018 11:29 pm  

we have about the same configuration except for the 1TB SSD, i went with 512GB. How is it working for you with the VEGA 64 in Mac?

2017 MBP 15″ RP560 + Mantiz Venus + AMD Radeon Pro VEGA Frontier Edition


ReplyQuote
Page 2 / 2 Prev