MacInTouch Amazon link...
Channels
Products
How about trying the "pop the bottom” . . .. Enquiring minds want to know!
Since the 2012 Mac Mini came to live here (that’s Brisbane, Australia - hot humidity, come summer, like now), it has sat on its side with the tin-can lid leaning on the ‘top’ side. When I feel the heat, a small personal fan arrives at the desk and blows not on me (I’m tough) but straight into the Mini.

No measurements to offer, it just runs cool.
 


Since the 2012 Mac Mini came to live here (that’s Brisbane, Australia - hot humidity, come summer, like now), it has sat on its side with the tin-can lid leaning on the ‘top’ side. When I feel the heat, a small personal fan arrives at the desk and blows not on me (I’m tough) but straight into the Mini. No measurements to offer, it just runs cool.
The "n" speed Airport Extreme I used for years ran hot and seemed to run cooler when set on its side. Had the original Apple TV with 40GB hard drive, and that ran so hot even at idle, I tested it with a Kill-a-Watt. Since it didn't have a native on/off switch, I attached it to a power bar to drive a stake in its power vampire heart. Had the "tower" Airport Extreme "ac" at work, and it died early, just after warranty expired. Google turns up a lot of "heat death" links about that device, so we moved on to a Netgear that has lots of holes for ventilation. That was before Apple left the router business.

Apple's "fix" for the i9 MacBook Pro did improve its performance but didn't address its inability to shed enough heat to avoid throttling when the GPU is active. Seems the 2018 Mini is another piece of Apple kit that is frustrating because it could do better, if designed to shed heat.

After Jen's post, I still wonder if the 2018 Mini with popped bottom and external fan would perform closer to capability.
 



David - I was one, among others, who suggested removing the base might be counter-productive if the fan design depends on the bottom seal to direct air flow. There's no real way to tell without trying. Sure wouldn't want it to be damaged in a "science experiment."
I found a description by "Sebastian" on eGPU Forums of his DIY cooling base. He took a 5V PC fan, mounted it in a wood base behind a filter, then set the Mini in sculpted foam to control air flow.
It is not clear whether he removed the Mini's OEM base, but from the photos it looks like he didn't.
George, I thought about that. Sebastion's setup gained him roughly a 5% gain in real life rendering speed. Seems like an awful lot of work for that. It didn't solve the problem, just made it slightly less intrusive, at the cost of three hours of work plus the parts.

At that point... if you can save more than three hours of work that way, you're better off with a different computer.

Taking the bottom off is on my agenda at some point. I truly don't think it'll work, because, as you yourself said, the fan may well depend on unit sealing. Indeed, the big Mac Pro I need to sell somehow was designed to only work with the case sealed; open the case, and it automatically shuts down.
 


So is it time to take a Dremel tool - carefully - to the 2018 Mac Mini, particularly the i7, to get the ventilation required to keep it from throttling?
I've discussed a similar idea with hardware engineers at a very high-end industrial electronics manufacturer, and I strongly advise against any case modification that would change internal air flow on a Mini or any other tightly-packed piece of hardware. Apple almost certainly modeled active and passive internal air flow extensively when designing the machine. If you add ventilation holes, you may end up increasing air flow to some parts of the machine while inadvertently decreasing air flow to other parts of the machine. It may be counterintuitive, but there is a very real chance that adding ventilation may make the heating problem worse!
 


It may be counterintuitive, but there is a very real chance that adding ventilation may make the heating problem worse!
That was my assumption. I think the key issue here is the processor itself. The only way around that is some heavy-duty cooling, e.g. liquid refrigerant or crazy-large active heat sinks. I really wonder what the plan is on the Mac Pro, but my guess is it will stick with the Xeon line and some sort of semi-solution will be built in.
 


I commented to Ric, perhaps several years ago now, that the current [2013] Mac Pro was actually the perfect case design for an updated Mac Mini. The thermal efficiencies of the case were already known, with much of the design and manufacturing equipment costs already off the books. It would have allowed for the hotter CPUs being used now, along with a mild discrete GPU if desired, perhaps even a spot to install a large 2.5" SATA III hard disk drive, in addition to anything Apple wanted to solder onto the logic board. The case would easily support the number of ports we see in the 2018 Mac Mini design. Paint the thing white, or space gray, and you are ready to go.

In my opinion, many of the problems and complaints associated with the current Mac Pro and Mac Mini would have evaporated in such a design. The downside would have been that it was not possible to rackmount these in a way to obtain the density we see in use at places such as MacStadium.
 


The Unofficial Pop The Bottom Off A Mac Mini And Aim A Fan Inside Unscientific Thermal Test

We have a variety of Mac Minis at work, several mid-2010 Core 2 Duos, a couple of 2012 Quad i7s, and three of the base 2014's. I decided to test the hypothesis that removing the bottom rubbery seal from the Mini case would enable it to run cooler, and to further test the effect of aiming a fan at the exposed innards.

Unfortunately our little herd of Minis doesn't include the newest 2018 models, and we disconnected the 2012 i7 in the office where I work from the Internet to avoid the possibility an update might disable expensive graphics printers. So I decided to test a mid-2010 with the following specs:
  • 2.4GHz Core 2 Duo
  • 8 GB 1067MHz DDR3
  • NVIDIA GeForce 320M 256MB
  • aftermarket OWC-installed 240GB SSD
  • macOS Sierra 10.12.6
Looking for a way to stress this venerable Mac, I decided to run simultaneous YouTube videos in Firefox and Chrome, each on its own monitor. The computer wasn't able to run two 1080p videos simultaneously, so I dropped the settings to 720p in each browser, and adjusted the speed settings as high as the computer would actually display moving video, however janky.

In the initial run, with bottom on as from factory, the system seemed to max out at 68.3 C (154.94 F) with a fan speed of 2771 RPM out of a possible 5000.

I shut down the Mini, and aimed a fan at it to cool the case. Coming back an unscientific time later, but long enough that it felt cool to the touch, I removed the bottom rubbery base. When rebooted, the Mini CPU reported 34.77 C, which at 94.6 F is a lot hotter than our office.

In the second run, I simply turned the Mini over, bottom off, on the standard computer desk (laminate surface on dense particleboard), and launched the same videos. Checking in on it after another unscientific interval, I could feel the heat rising from the exposed interior. Again, after an unscientific period, I checked and it had risen to 63 C, though the fan speed was only 1807, barely over the minimum speed of 1800. In the early stage of this run, I didn't notice the temp rising as fast as it had when the Mini was "closed," but when I ended the run the top of the Mini case that was sitting on the laminate was "heating pad" warm to the touch, as was the desk surface on which it rested. I presume the room temperature desk had, however inefficiently, been acting as a heat sink, until both the Mini and the surface in which it was in contact grew hot. It is possible that had I set the Mini not on the desk surface but on a ventilated rack it would have run cooler.

Last run I set the Mini on its side, aimed a fan at its exposed interior, and beginning from the 63 C it had reached in run 2, watched (and felt with my hands) as its temps and case cooled. I'm ending my experiment with the Mini at 55.4+ C and fanspeed 1800. It has been in that range running long videos quite some time. That's 131.72 F, 13 degrees C and 23.22 F lower than with the case sealed. About 20% cooler -

Another very unscientific observation: Set on its side with the fan blowing into its interior, the videos seemed to play with less stutter.

I'm pretty sure this is the first time since the Mini, purchased refurbished from Apple in 2011 and shipped directly to OWC for the SSD install, has been opened since OWC worked on it. The interior was remarkably clean, the fan shows no easily visible dust build up, and curiously the only dust I found was a lonely "dust bunny" inhabiting a "compartment/divider" in the rubbery base.

I used the freeware Fanny "widget" to monitor temps and fan speed.
 


Ric Ford

MacInTouch
Looking for a way to stress this venerable Mac, I decided to run simultaneous YouTube videos in Firefox and Chrome, each on its own monitor.
I'm not sure how consistent and reliable YouTube streaming would be across a variety of tests - in other words, I wonder if network effects/variations might overwhelm hardware effects/variations. It might be a better test to run the videos off a local drive.
 


I'm not sure how consistent and reliable YouTube streaming would be across a variety of tests - in other words, I wonder if network effects/variations might overwhelm hardware effects/variations. It might be a better test to run the videos off a local drive.
My goal wasn't exactitude, but to heat the Mini up by running it hard, then see if it ran cooler without the base cover and / or assisted by an external fan.

We've wondered in this thread if the new Mini's cooling design depends on the sealed bottom. The 2010 Mini I tested doesn't, and its fan speed as reported by the "Fanny" app wasn't "over-pushed" by lively exterior air input.

One observation I didn't mention earlier: when the bottomless Mini was upside down on the desk (no exterior fan), and I felt heat rising from the interior, I also checked the fan's exhaust vent, which was blowing out more heat than was rising from the interior. The fan was working effectively without the base.

If the 2018 Mini could shed anything like the 20% of CPU heat I observed with a bottomless, exterior-fan-assisted 2010 Mini, it shouldn't throttle, and its hardware life expectancy should be longer. I doubt most users will care, or would want to rig a protective filter and effective fan. Others, who are looking to use the Mini's strong CPU for the kind of intensive applications that will induce throttling, might.
 


I keep seeing people talking about thermal limits and throttling with the i7 - under actual loads, I saw it throttle back for less than 1 minute while the fan ramped up, and once the fan ramped up, it was getting as good or better performance to when it was cold.

Ran different tests and workloads several times, once I read about people seeing major throttling - I could not reproduce with the i7 or 2x i5 models I have here (and the i3 stays cool under most loads I put it under - using nearly 1/3 the electrical power of the i7 during most tasks - idle is the only point where they are close, with the i3 using about 2/3 of the idle power of the i7).

As for popping the rubber off - the 2014 and 2018 Mac Minis have an extra shield on the bottom with many ventilation holes (check the iFixit teardown for images), so should be safe to remove without endangering the internals - maybe prop the Mini up a bit from the surface to improve airflow if the surface is solid - something rubbery to prevent vibrations as well.

I may re-run my more extensive 2018 i7 tests with the bottom off to see if it makes any difference.

On my 2012 quad-i7 at home, I prop it up so it's a couple inches above the wood desk surface. I had an external fan set up at one point to blow across the case, and it did help - but was a bit noisy and was more cluttered. I found that using one of the fan control apps to just set a higher minimum speed covered most of my needs.

I do remember seeing one of the Mac-based hosting companies that has hundreds of Mac Minis having them all on their sides.

There are some nice USB fan models out now with variable speed control and such things that could be considered - like the ThermalTake Mobile Fan II (or III which has a standard USB connection and slightly nicer case).
 


I'm not sure how consistent and reliable YouTube streaming would be across a variety of tests - in other words, I wonder if network effects/variations might overwhelm hardware effects/variations. It might be a better test to run the videos off a local drive.
Here's a link to an excellent way to stress-test a Mac using Terminal:
Stress Test a Mac by Maxing Out CPU
yes > /dev/null & yes > /dev/null & yes > /dev/null & yes > /dev/null &
I'm using it to test my 2011 Mac Mini Server. Results will be posted shortly.
 


...I strongly advise against any case modification that would change internal air flow on a Mini or any other tightly-packed piece of hardware. Apple almost certainly modeled active and passive internal air flow extensively when designing the machine. If you add ventilation holes, you may end up increasing air flow to some parts of the machine while inadvertently decreasing air flow to other parts of the machine. It may be counterintuitive, but there is a very real chance that adding ventilation may make the heating problem worse!
Careful. This is the same company that shipped the original Apple Base Station without any ventilation holes whatsoever. The units I modified definitely had a longer life than the ones I didn't. That's not to say that Apple didn't model the daylights out of each design, but the current issue at Apple is the triumph of design over functionality.

Never underestimate the desire by Sir Ive and his merry band of fellow design zealots to push the limits of all things engineering just to get the courageous "look" that they desire. The trash can Mac Pro is a perfect example - a unit that could barely handle the thermal envelopes of extant CPUs/GPUs of the day when pushed and hence never subsequently updated.

Even today, long-discontinued cheesegrater-era cases can be upgraded to use more powerful CPUs and GPUs than currently shipping products. Why? For starters, a better power supply, better inherent air flow, and greater room for expansion. The trashcan Mac Pro (like the G4 cube before it) is an homage to whatever design aesthetic ruled that day at Apple, engineering consequences be damned.

A better "Pro" design decision at Apple would have been to further refine the cheesegrater platform, as Apple has yet to ship a better "pro" case re: air flow and versatility. It's one reason that cheese graters continue to enjoy the popularity they do. Apple could likely print a mint by releasing motherboard upgrades for the cheesegrater series as an interim stepping stone towards a new platform in the future...

... that is if they actually cared about the "pro" market.

In closing, I see nothing wrong with experimenting. However, the risk vs. the reward may be low, depending on the use case. Putting the CPU in a temperature-controlled space may be cheaper than modifying the case.
 


Here's a link to an excellent way to stress-test a Mac using Terminal...
I'm using it to test my 2011 Mac Mini Server. Results will be posted shortly.
In my search for reports from users who removed the base from Mac Minis, I ran into suggestions to use this terminal command to stress-test CPUs - and several warnings that if there are issues with a Mac, be it bad solder or the infamously craptacular GPUs that are killed by heat failure, this test run too long on too many cores will find that weakness. Just a heads-up. Not that other company's systems are immune, but...
 


ASUS (press release) said:
ASUS Announces New Mini PC ProArt Series for Professional Content Creation
For content creators who deal with heavy design workloads daily, system stability and a high-performance cooling system is as essential as a powerful processor.
ProArt dimensions: 6.9 x 6.9" x 14.4" tall - includes internal Nvidia Quadro workstation-grade graphics.

Mac Mini dimensions: 7.7 x 7.7" x 1.4" tall + BlackMagic eGPU Pro 6.96 x 6.96" x 11.59" tall.

This is the interesting part most relevant to our ongoing discussion of Mac Mini cooling:

Mini PC ProArt PA90 is equipped with an all-in-one CPU liquid cooler that unleashes the full performance of the CPU while delivering consistently quiet and stable operation. The liquid cooler also enables a quieter working environment, operating at just 32dB at full CPU load, 36% quieter than average desktop PCs.
An innovative ventilation system also helps ensure cool and stable performance. When the CPU temperature rises above 80°C, the top cover of Mini PC ProArt PA90 automatically extends, allowing 38% more airflow into the chassis for optimized thermal efficiency.

Additional specs, though no price information, at press release link above. Noteworthy that the system ships with options for Windows 10 or "no operating system," implying it is Linux-compatible.
 


... I do remember seeing one of the Mac-based hosting companies that has hundreds of Mac Minis having them all on their sides.
Assume that those Minis are in an environmentally-controlled room. Someone putting this in their unheated/uncooled garage, or in a home office desk, or under a monitor stand, might see throttling more....

Ah, some good news! My desktop 2012 Mac Mini is now being replaced with a 2018 MacBook Pro. Looks like I might get chance to mod/upgrade the Mac Mini to SSD... wish me luck!
 


Here's a link to an excellent way to stress-test a Mac using Terminal ... I'm using it to test my 2011 Mac Mini Server. Results will be posted shortly.
Here are the stress results for my 2011 Mini server with an Intel Core i7-2635QM processor (4 cores, base frequency 2 GHz, max Turbo 2.9 GHz).

I used the terminal commands referenced above to ramp the processor utilization up to 100%. Each step took about 6 minutes to stabilize, with the total time about 40 minutes. The computer is on a shelf in my garage with an ambient temperature of ~72 F. I used Intel Power Gadget and Macs Fan Control to take the data.
Code:
Utilization %    Freq. GHz     Fan Speed     Temp. °C
   0.7              0.8            2300        64
  12.6             2.64            4397        89.8
  25.0             2.60            5492        98
  37.3             2.37            5517        96.8
  50.3             2.16            5495        97.4
  74.5             2.10            5495        97.1
  98.8             2.06            5517        97.3
   0.8             0.86            2300        63.7
It looks like the cooling system is able to maintain its rated CPU frequency (2 GHz) pretty continuously while enabling significant turbo usage to somewhat above 25% utilization.
A few points about this test:
1) It's processor only, no significant hard drive or GPU usage.​
2) The fan ramps up first, until the temperature approaches ~100 C, the chip limit.​
At that point, as the utilization is upped, the frequency drops to maintain a​
temperature slightly below 100 C.​
3) I noticed the individual core temperatures were changing, indicating that the processing​
power was switching from one core to another at the lower two levels.​
 


Ric Ford

MacInTouch
Here's a link to an excellent way to stress-test a Mac using Terminal...
It looks like this technique was already being used by 2006, if not earlier, for Mac hardware assessment:
Jason D. O'Grady said:
Test for MacBook Random Shutdown Syndrome (RSS)
This is only a test. MacBook Random Shutdown Syndrome seems to becoming more common with MacBook owners. Luckily there's a simple test to diagnose if your MacBook is afflicted.

Ash from Techpaedia.com was annoyed by his new MacBook shutting down randomly so he developed a simple test to see if a MacBook suffers from Random Shutdown Syndrome (RSS). To see if your MacBook has RSS:

1. Open Terminal​
2. Type the following command and hit return: yes >/dev/null &
3. Repeat step #2 again in Terminal: yes >/dev/null &
Wikipedia has more about the yes (Unix) command.
 


Just imagine how nice the Mac Mini could have been if it had been made three times the height so it could have accepted four DIMMs and two 2.5" SSDs. The case could then have been properly ventilated so the CPU and GPU would not have to be throttled.

But as long as there are those who drink Ive's Kool-Aid and buy overpriced, under-engineered cutesy fluff, they'll have to live with it. ...
 


Here's a link to an excellent way to stress-test a Mac using Terminal:
Stress Test a Mac by Maxing Out CPU
yes > /dev/null & yes > /dev/null & yes > /dev/null & yes > /dev/null &

I'm using it to test my 2011 Mac Mini Server. Results will be posted shortly.
Here are the stress results for my 2011 Mini server with an Intel Core i7-2635QM processor (4 cores, base frequency 2 GHz, max Turbo 2.9 GHz).

I used the terminal commands referenced above to ramp the processor utilization up to 100%. Each step took about 6 minutes to stabilize, with the total time about 40 minutes. The computer is on a shelf in my garage with an ambient temperature of ~72 F. I used Intel Power Gadget and Macs Fan Control to take the data.
Code:
Utilization %    Freq. GHz     Fan Speed     Temp. °C
   0.7              0.8            2300        64
  12.6             2.64            4397        89.8
  25.0             2.60            5492        98
  37.3             2.37            5517        96.8
  50.3             2.16            5495        97.4
  74.5             2.10            5495        97.1
  98.8             2.06            5517        97.3
   0.8             0.86            2300        63.7
It looks like the cooling system is able to maintain its rated CPU frequency (2 GHz) pretty continuously while enabling significant turbo usage to somewhat above 25% utilization.
A few points about this test:
1) It's processor only, no significant hard drive or GPU usage.​
2) The fan ramps up first, until the temperature approaches ~100 C, the chip limit.​
At that point, as the utilization is upped, the frequency drops to maintain a​
temperature slightly below 100 C.​
3) I noticed the individual core temperatures were changing, indicating that the processing​
power was switching from one core to another at the lower two levels.​
Does this test show anything more than the system behaved as it should, ramping up fans and lowering GHz as the chip reached its 100C "T Junction?"

I'm also not sure how to read the results in TomJ's table.
Intel said:
Intel® Core™ i7-2635QM Processor
  • # of Cores 4
  • #of Threads 8
  • Max Turbo Frequency 2.90 GHz
  • Intel® Hyper-Threading Technology (Intel® HT Technology) delivers two processing threads per physical core. Highly threaded applications can get more work done in parallel, completing tasks sooner.
In the table, the reported processor speed doesn't reach the 2.90 GHz "Turbo," apparently topping out at 2.64 GHz, which is substantially higher than its 2.0 GHz "Base," but lower than 2.90?

In the 2012 OSXDaily link, neither the Mac being stressed, nor its processor, are revealed. From the accompanying video, the command was (best I could tell) invoked 18 times. It was my understanding that to fully "stress" a CPU, the command needs to be invoked at least once for each physical core and once for each virtual one. Not sure why it was necessary to invoke the command 18 times. Certainty? Or testing on the dual 6-core Xeon Mac Pro? TomJ72: how many times did you invoke the command?

From what I've been reading, Intel's hyperthreading provides the attack vector for the much-publicized Spectre and Meltdown. Apple released some mitigations for Spectre/Meltdown beginning in January, 2018. Effects on hyperthreading?
 


A really simple way to stress-test a CPU with a multi-threaded assault is to use Handbrake to modify a large video file.

For example, I found out that the alleged 60W power delivered by a external Thunderbolt 3 dock is not quite enough to cover all the power needs of a Touchbar MacBook Pro 15" while converting 1080p video at 2fps. It took almost a day, but the power in the battery steadily dipped until it was below 30%.

Anyhow, Handbrake is free and works across many platforms and operating systems.
 


In the table, the reported processor speed doesn't reach the 2.90 GHz "Turbo," apparently topping out at 2.64 GHz, which is substantially higher than its 2.0 GHz "Base," but lower than 2.90?
This is not necessarily a problem. The "turbo" speeds are only achievable if there is sufficient software load to require it and only if the thermal solution is able to keep the CPU from overheating at that speed.

If the CPU is loaded and not reaching that maximum frequency, it probably means that the existing cooling solution can't keep the temperatures low enough, so the turbo boost firmware limits the frequency to the highest it can go without overheating.

I wouldn't consider this a design fault, since I don't think it is reasonable to expect these maximum clock speeds to be reached for more than short periods of time even under ideal conditions. In my opinion, if the system is always running at or above the base frequency (except when idle), then it is running within specifications and is operating normally.

In other words, just because there's a turbo boost frequency limit of 2.9GHz, it is still a 2GHz processor and should be treated as such.
 


This is not necessarily a problem. The "turbo" speeds are only achievable if there is sufficient software load to require it and only if the thermal solution is able to keep the CPU from overheating at that speed....In other words, just because there's a turbo boost frequency limit of 2.9GHz, it is still a 2GHz processor and should be treated as such.
Agreed, but only to a point. If that period of time that a CPU can maintain 2.9GHz can be measured in nanoseconds, then I'd argue that the 2.9GHz clock speed is irrelevant... and hence shouldn't be advertised. After all, customers are potentially paying to upgrade between processors with varying speeds, core counts, and so on.

Given how many factors come together to determine what the maximum single-core clock speed can be, I also recognize the difficulty of coming up with a meaningful reference/test procedure. Perhaps Intel has published a specification?

I recently added a fan and shroud to a CPU under my care, along with a shim for better CPU die contact to the heat sink, and better heat transfer paste. The result: an idle die temperature just 2°C over ambient and 20°C lower than the previous idle temperature. This CPU likely can handle longer bouts of "Turbo" performance and/or enjoy a longer life than before its cooling system underwent these modifications.

Good luck doing something similar inside a Mac Mini, iMac, Mac Pro, or pretty much any Mac at this point. There is no room, no use of standardized parts, etc. Your likely best bet is to either duct additional air via a fitted external blower system and/or to place the unit in a controlled (cold) environment.

That's one benefit of the "PC" universe: the ability to choose among a wide / bewildering range of motherboards, accessories, cases, power supplies, etc. The resultant systems reflect the needs/wants/preferences of the builder in ways that Apple hardware today simply cannot, because Apple is no longer a computer company that caters to a wide range of computer customers. Nowadays, Apple has become a phone manufacturer that happens to make computer systems on the side, systems that increasingly reflect the ongoing influence of iOS and the iPhone (APFS, glued-in batteries, RAM and flash soldered on to the motherboard, touch bar, T2 chip, fingerprint reader, etc.).

I'd treat any performance claims from Apple or any other manufacturer with a grain of salt. Trust, but verify with a representative work load, if your professional life depends on it. Also, make sure that any machines with fans get serviced every few years if your work environment is dusty (for example, lots of humans around), because dust on heat exchangers will impose a toll on the cooling system performance, just like it does on the home refrigerator.
 


Ric Ford

MacInTouch
Whoa... this is not cool - another Apple design defect or software bug?
OWC said:
Envoy Pro EX
Bus-powered Thunderbolt 3 solutions like the Envoy Pro EX may experience intermittent disconnects when attached directly to one of the Thunderbolt 3 ports on the Mac mini. To avoid this issue, we recommend connecting your Envoy Pro EX through another wall-powered Thunderbolt 3 device in a daisy-chain. These intermittent disconnects only occur when connected directly to the Mac mini (2018) and are not seen on other computer models. This issue has been reported to Apple for resolution.
 



I'm also not sure how to read the results in TomJ's table.
In the table, the reported processor speed doesn't reach the 2.90 GHz "Turbo," apparently topping out at 2.64 GHz, which is substantially higher than its 2.0 GHz "Base," but lower than 2.90?
I’m also not 100% sure how to read the table. It’s worth noting a few things, though.

Intel’s maximum Turbo mode frequencies vary based on a lot of factors. The simplest is the number of active cores. For the i7-2365QM, if only a single core is active, it can run at up to 2.9 GHz (potentially limited by power and cooling). Using a second core drops the Turbo for both to 2.8 GHz. Touching the third or fourth core drops Turbo for all to 2.6 GHz. (This is sometimes called the “all-core turbo” frequency.) There’s often a lot of background activity on a Unix (including OS X) system, as Activity Monitor shows. This will reduce the amount of time that a single-core, or even dual-core, turbo mode is active.

This processor steps its cores by 100 MHz at a time, so 2.0 through 2.9 GHz are possible, but any measurements which aren’t exactly on a 100 MHz boundary are likely averages across either multiple cores or time. (The general rule for Intel processors is that they run at integer multiples of their bus speed.)

Using “yes” isn’t a very stressful test of a processor, because it’s purely using the integer arithmetic portion of the processor, which draws the lowest power. Handbrake is a much better system stress test, as it exercises the floating-point and vector engines as well. (It’s trivial to write a small program which can be used for per-core stress testing too, using [floating-point] and vector arithmetic.) In other words, trying to determine if the maximum speed of a processor is going to be overly constrained by power or thermals requires a benchmark application whose instruction mix is similar to that which you care about.

The integrated GPU shares the same power supply, so using it for transcoding or 3D graphics will also reduce the amount of power available to the CPU, and reduce the maximum Turbo frequency and the amount of time the processor can stay in Turbo.
 


Does this test show anything more than the system behaved as it should, ramping up fans and lowering GHz as the chip reached its 100C "T Junction?" I'm also not sure how to read the results in TomJ's table. In the table, the reported processor speed doesn't reach the 2.90 GHz "Turbo," apparently topping out at 2.64 GHz, which is substantially higher than its 2.0 GHz "Base," but lower than 2.90? In the 2012 OSXDaily link, neither the Mac being stressed, nor its processor, are revealed. From the accompanying video, the command was (best I could tell) invoked 18 times. It was my understanding that to fully "stress" a CPU, the command needs to be invoked at least once for each physical core and once for each virtual one. Not sure why it was necessary to invoke the command 18 times. Certainty? Or testing on the dual 6-core Xeon Mac Pro? TomJ72: how many times did you invoke the command?

From what I've been reading, Intel's hyperthreading provides the attack vector for the much-publicized Spectre and Meltdown. Apple released some mitigations for Spectre/Meltdown beginning in January, 2018. Effects on hyperthreading?
Yes to get full CPU usage on the 2011 Mini Server, you need to invoke the "yes" command 8 times. I did the test to see what the old Mini would do.

As Constantin points out, many graphics programs, like Handbrake, can utilize almost the entire capability of the computer.

My test shows, if not stressed too much (25% utilization, 2 "yes" commands), you can get higher frequency in the CPU, and at full utilization, the computer seems perfectly happy at its rated CPU frequency (2 GHz). No, it never reached 2.9 GHz, but I would suspect there's probably a limit in the board for safety sake. So thermally this old computer, after running in the garage for over three years, is still up to snuff with all the dust and bugs that may have crawled in. (No roaches, however - our South Carolina roaches are too big to get in a Mini.)
 


Yes to get full cpu usage on the 2011 mini server you need to invoke the "yes" command 8 times. I did the test to see what the old mini would do. As Constantin points out, many graphics programs like Handbrake can utilize almost the entire capability of the computer. My test shows if not stressed too much (25% utilization, 2 "yes" commands) you can get higher frequency in the cpu and at full utilization the computer seems perfectly happy at it's rated cpu frequency (2 GHz). No, it never reached 2.9 GHz, but I would suspect there's probably a limit in the board for safety sake. So thermally this old computer after running in the garage for over three years is still up to snuff with all the dust and bugs that may have crawled in. No roaches however, our South Carolina roaches are too big to get in a mini.
A couple of things I forgot to emphasize:
1) When running at 12.5% or 25% utilization, the computer is not using a single virtual CPU but is periodically switching from one core to another. This is shown by the temperature variations shown in each core by Macs Fan Control while running at 12.5% or 25%. Above 25% the variations were within the noise level of the test.​
2) I noticed, for my 2011 Mini Server and my 2012 Mini Server, the Intel spec sheets list the temperature limits for the cores themselves. For the i3 in my hackintosh, the temperature limit is listed for the case of the CPU.​
 


The Unofficial Pop The Bottom Off A Mac Mini And Aim A Fan Inside Unscientific Thermal Test...
Maybe if the bottom cover is removed and this is placed underneath the Mac Mini, it might provide better cooling. Sure, an adapter plate will be needed for the Mac Mini to sit upon. Should be easy to make.
Just a crazy idea.
I did just that with the AC Infinity referenced above, repeating the last step of the three-stage test I ran earlier. This time I set the bottomless Mini on the Multifan's nice suction cup-looking rubber vibration dampers (there are four on top, four on bottom) and ran it just short of an hour with "endless videos."

It ended at 58.74°C with fan speed at 1802, barely over the minimum 1800 at boot. The previous test ended with with the Mini at 55.4+°C and fanspeed 1800. That was 13° C and 23.22° F or about 20% cooler than with the case sealed.

Unlike when I had a more powerful "table fan" aimed at the Mini set sideways, I did notice the top of the case was warm to my touch when I shut it down. Given how inexpensive these fans are, I suppose an additional Multifan sitting on top could lower the case temperature. The Multifan is very quiet.
 


Thank you, George, for the test and results. Good results there. Quite interesting. Maybe for the top of the Mac Mini, a heatsink might work.
 


Thank you, George, for the test and results. Good results there. Quite interesting. Maybe for the top of the Mac Mini, a heatsink might work.
Don't know how efficiently, but the Mini's aluminum case dissipates heat. I found one FLIR (thermograph) image of a 2018 i3 being being benchmarked, and it was yellow/orange all over.

Somehow attaching to the Mini's case the kind of CPU cooler PC overclockers use would pull heat off the case. But if the goal is to load the 2018 Mini and keep it below its throttling threshhold, the inexpenive MultiFan you suggested and I tested today could be enough. A lot better desktop companion than the 12" table fan I tried last week, and nearly as effective.
 


Don't know how efficiently, but the Mini's aluminum case dissipates heat. I found one FLIR (thermograph) image of a 2018 i3 being being benchmarked, and it was yellow/orange all over.
Somehow attaching to the Mini's case the kind of CPU cooler PC overclockers use would pull heat off the case. But if the goal is to load the 2018 Mini and keep it below its throttling threshhold, the inexpensive MultiFan you suggested and I tested today could be enough. A lot better desktop companion than the 12" table fan I tried last week, and nearly as effective.
I use my 2012 Mini I7 quad for photography, and Macs Fan Control does a good job of responding to every tick up in processor temp. Still, when running a batch for 30-60 minutes, the fan can not keep the heat down when going full blast. In my freezer I have some gel packs in a 'sock' left over from some physical therapy application. I plop one on top of the Mini, and the temp drops to normal using the heat dissipation of the case, and the fan is quiet. So, a freezer is not the typical computer accessory, but it can be handy.
 


I use my 2012 Mini I7 quad for photography, and Macs Fan Control does a good job of responding to every tick up in processor temp. Still, when running a batch for 30-60 minutes, the fan can not keep the heat down when going full blast. In my freezer I have some gel packs in a 'sock' left over from some physical therapy application. I plop one on top of the Mini, and the temp drops to normal using the heat dissipation of the case, and the fan is quiet. So, a freezer is not the typical computer accessory, but it can be handy.
Ingenious!

Two concerns:
  • Water condensing on the underside of the aluminum case?
  • Large temperature differentials stressing solders and plastics?
I presume high fan volume will blow vapor out the case before it can form water drops. Not sure about the "sudden" effect on components of plopping the ice pack on an already hot Mini, but, again, I presume the case and internal temperature change wouldn't actually be sudden as the Mini's heat won't move into the gel instantly, and the busy fan should moderate the sudden effect.

Over in PC Land, there are stories like this:
maketecheasier said:
4 Amazing Computer Cooling Techniques and Their Flaws
There are, however, a few people who aren’t satisfied with simply thinking outside the box when it comes to CPU/GPU cooling. They take that box, rip it out, and put their own in its place. Despite their best efforts, there’s still a good reason why we’re sticking to heat sinks and fans.
 


I have been following this discussion since the 2018 Mac Mini was released.

I am using a Mac Pro 3,1 with an SSD on an OWC PCIe card, My user account is on one of my hard drives. I basically like it. However, many programs that I like are coming out requiring macOS 10.12 or higher. I don’t know what the new Mac Pro will provide, but I expect it to be very expensive. So I thought the Mac Mini would provide an option. I would network it to my Mac tower to make use of its hard drives and optical disk drives. But this discussion scares me. Overheating really seems to be a problem. My daughter’s laptop died because the GPU overheated and died.

So my questions are:
  • What programs actually make use of multiple cores?
  • Is there a way to monitor which core is doing what?
  • What programs actually make use of a real GPU - excluding games?
  • What is the impact on the Mac Mini temp when it is powering various accessories?
 


Ric Ford

MacInTouch
But this discussion scares me. Overheating really seems to be a problem. My daughter’s laptop died because the GPU overheated and died. So my questions are:
  • What programs actually make use of multiple cores?
  • Is there a way to monitor which core is doing what?
  • What programs actually make use of a real GPU - excluding games?
  • What is the impact on the Mac Mini temp when it is powering various accessories?
I don't think you should be scared of the Mac Mini having inherent thermal problems. While I was testing the high-end version, it worked just fine and handled the heat without a problem and very quietly. That said, if you get one with defective thermal paste, like SnazzyLabs did, that's a different story - a quality problem, rather than a design problem. And if you're going to be transcoding video all day every day, then, no, it's probably not a good option. But for regular work, it should be very nice.

Regarding the use of multiple cores, this is growing every day and already in widespread use, including macOS/Finder and other Apple software. My experience with the 6-core i7 Mini was very good, better than with my 4-core i7 systems - more "fluid", and I'm actually missing that a bit.

Activity Monitor works well to display both CPU/core usage and GPU usage (in Mojave). Graphics software (e.g. digital photo editors) should make full use of a real GPU - as it did with the eGPU I set up. Text editors? Probably not. Video? Certainly.

I didn't see any impact on Mac Mini temps from accessories/peripherals, including external high-speed SSDs or the eGPU or Thunderbolt Target Disk Mode.
 


Graphics software (e.g. digital photo editors) should make full use of a real GPU - as it did with the eGPU I set up. Text editors? Probably not. Video? Certainly.
In my case, there's no question but that Photoshop is faster under the “slow GPU, fast CPU” Mac Mini than under the “slow CPU, fast GPU” Mac Pro (even with the six-core Xeon update).
 


Ric Ford

MacInTouch
In my case, there's no question but that Photoshop is faster under the “slow GPU, fast CPU” Mac Mini than under the “slow CPU, fast GPU” Mac Pro (even with the six-core Xeon update).
Do you have benchmarks (e.g. Geekbench "Compute") results to compare the two GPUs? With the rapid improvements in integrated GPUs in recent years, the Mac Mini's isn't slow by old standards.

And what Photoshop operations are you talking about, exactly, that are faster on the Mac Mini vs. the Mac Pro?

Also, I can't remember if you're running Mojave, but if you are, can you check GPU (and CPU core) usage in Activity Monitor?
 


But this discussion scares me. Overheating really seems to be a problem. My daughter’s laptop died because the GPU overheated and died.
I don't think you should be scared of the Mac Mini having inherent thermal problems. While I was testing the high-end version, it worked just fine and handled the heat without a problem and very quietly. That said, if you get one with defective thermal paste, like SnazzyLabs did, that's a different story - a quality problem, rather than a design problem.
Also note that modern computers (Macs and PCs alike) implement thermal throttling. A chip is not going to run so hot that it burns up and fails. Before it gets to that point, internal sensors will force it to slow down. In other words, a thermal problem will manifest as a performance problem, not as a system completely dying.

And, as has been reported here, Apple's new Mac Minis don't exhibit any such problems under normal use, and even under stress continue working fine (but with loud[er] fan noise).

All the talk here about alternative cooling solutions is really focusing on how to get the maximum possible performance out of the CPU chip when it is under extreme stress, not compensating for a design defect.
 


Ric Ford

MacInTouch
Also note that modern computers (Macs and PCs alike) implement thermal throttling. A chip is not going to run so hot that it burns up and fails.
A little historical perspective on that:
Apple Discussions said:
Overheating graphics card failing, need more cooling?
I have a macbookpro, 2.66 ghz intel core i7 with 8 gb ram. I have 2 video cards from what I can tell, that are listed under my system profiler: The Intel HD Graphics and a NVIDIA GeForce GT 330M. When I play world of warcraft on Bootcamp, my fans kick in after about 10min when the temperature gets high, but it is no longer enough to keep my graphics card from crashing. I get a bunch of colored squares streaking across my screen and the laptop either recovers right away, or has to force a restart. I am thinking that either my graphics card has finally had enough and can no longer handel the heat, or my macbook is no longer cooling properly. Are there any other ways of keeping it cool? My fans will run at full speed and I have a cooling stand, but neither are enough, it still gets too hot to touch.
MacWorld U.K. said:
MacBook Pro graphics failures addressed by Apple repair programme
Reports of failing MacBook Pros have been flooding in since 2013, with many owners of 2011 models with AMD graphics suffering from system crashes and hardware problems that have been described as "critical".

... On 19 February 2015 Apple announced a repair program for some MacBook Pro models suffering from the graphics issues we've described in this article.

It's not officially a replacement program, but instead is titled "MacBook Pro Repair Extension Program for Video Issues."

On its support page, Apple reveals it has determined "that a small percentage of MacBook Pro systems may exhibit distorted video, no video or unexpected system restarts".

The specific symptoms described by Apple include distorted or scrambled video on the computer screen, no video on the computer screen (or external display) even though the computer is on, and the computer restarts unexpectedly.
 



Amazon disclaimer:
As an Amazon Associate I earn from qualifying purchases.

Latest posts