Okay into the breach again new quad 2.3 arrives RAM tests and more.

[dasx]

Looks like normal. I also get around 100ºC when stessing the CPU for long periods of time. (Plus fans at 5500rpm).

That's why I plan on modding a spare bottom cover I bought. Won't be able to do it for some days. I'll post results as soon as I have it all done.

FYI, I noticed a 7ºC temp drop when moving my old 2010 Mac Mini to another more ventilated location. And some extra 8ºC when removing the bottom cover and elevating the mini a little bit with the edges of two books. That's why I wanna mod this extra bottom cover.

Originally Posted by MoMacFan

Hi Philipma and all others - thanks for covering this topic so thoroughly!

I am a new Mac Mini owner (I have a Macbook Air since about one year and needed some more serious computing power for rendering, transcoding etc.) and a bit worried of the temps I am reading with my new 2012 2.3 Quadcore Mac Mini, Fusion drive and stock Apple 8 GB Ram.

When running handbrake or prime95 (stress test with 8 "cores" at 100%) my CPU temps (1/2/A read with iStat) go up to 221 F/105 C and the fan spins at 5500 rpm.

I have tried to elevated my Mini and other tips I found here, but have seen no change in temperature whatsoever.
Since so far prime95 did not show me any error (around 4 hours into the stress test when I stop that in order not to kill anything at those temperatures).
So I am a bit concerned, especially when I see that with your new 2.3 quad Mini you will have your fan go up only to 3100 rpm (if I understood correctly) - leaving you with enough headroom - but mine is already maxed out at 5500 rpm. And as well it is no fun working besides a mini that runs so loud with the fan blasting away when under load (but when only browsing, running Office, software development etc. is is very quit and nice).

Have you done anything else in order to achieve these low rpm/temperature ratios?
Should I bring my Mini to the Apple store - even though technically it seems to work ok so far?
Should I go ahead and have prime95 run for a couple of days in order to make sure it is stable or would this be in the contrary contra-productive because it will accelerate the aging of the components?
Or should I further try find the cause for this, e. g. bad contact of the heatsink with the CPU and exchanging the thermal paste with Arctic Silver etc.?

Thanks,
Alex

[philipma1957]

Originally Posted by Snowcake

Edit: Did you also try to remove the black bottom and put the Mac Mini on its side? You should see a temperature drop if you do that.

try side mount and try this holder

http://eshop.macsales.com/item/NewerTech/NUSTALYMINI/

[MoMacFan]

Originally Posted by Snowcake

What is your room temperature? Philipma's room temperature is 27C.
Your Mac Mini is also running as hot as the faulty Mac Mini from Philipma.

I this happened to me: 5500rmp and the temperature running at TJunction(105c) i simply would replace it because if it can 3100rmp it should be like that on all Mac Mini's. You also lose warranty if you replace thermal paste by yourself.

Edit: Did you also try to remove the black bottom and put the Mac Mini on its side? You should see a temperature drop if you do that.

My room temperature is 20C (68F).
As you suggested I removed the black bottom and put the Mini on its side (power plug downwards): While running prime95 I can see now some changes (before it was just sitting at 105C and 5500rpm).
Within 5 minutes into the test the CPU 2 temp (that is the highest one, under load always ~5C above CPU 1) alternates between 90C and 99C (194F-210F) and the fan rpm is between 4700 and 5500, but more often above 5000.

Could the Fusion drive have some impact on temperature? At least the external SSD that I use with my Macbook Air is getting quite hot and maybe this is the case with the internal SSD (plus the 1TB Harddrive) as well?
That could explain differences with Philipma's Mini - if he has no Fusion drive, that is - other than a maybe sloppy job while attaching the heatsink in the factory or a defective CPU (but until now it passed all CPU-test)...
I might "torture" mine with prime95 over night to see if it is failing, it only hurts me a bit to do this to my nice little Mac (strange, I never had those thoughts with my PCs in all those years .-) ...

EDIT: With the bottom plate back on but still on its side it "behaves" again the same as horizontal (5500 rpm and TJunction)

[motrek]

Originally Posted by MoMacFan

...
Could the Fusion drive have some impact on temperature? At least the external SSD that I use with my Macbook Air is getting quite hot and maybe this is the case with the internal SSD (plus the 1TB Harddrive) as well?
...

Very odd that your SSD would be getting hot. Are you writing enormous amounts of data to it? Just reading data or letting the drive idle shouldn't cause it to heat up at all unless the USB controller chip isn't working right or is very poorly designed. I have several SSDs that I have used as external drives and they never get hot, even when writing.

[propower]

Just to add to the online data:

2.6 MAcMini Apple 256SSD 16G ram
I took the load to 100% on all processors (yes dev/null test)
Mini is elevated on 1" tall 4" square woodblock with 80mm 18dBA 1200 rpm fan blowing on back left corner.

Temps stabilized with 5500 rpm internal fan between 92degC and 96degC.

For a steady 22% load: one dev null instance and 2200rpm fan minimum via istat CPUA stabilized at 83degC

[MoMacFan]

@propower - for the comparison:
Doing those load tests (yes dev/null) with all 8 hyperthreaded cores of my 2.3 quad Mini, I reach 94C of CPU A at 5200rpm.
However using prime95 is far more taxing and my system reaches 105C.

@motrek:
It was just a hypothesis - the external SSD I own and that is getting nicely warm is a 256Gig Thunderbolt drive with passive cooling (aluminum or magnesium body).

@dasx:
So it could be "normal"? At least the Mac Mini I have is no server model, so it might be normal and might explain why the official server Mini comes with a lower clock and without fusion drive...

[motrek]

Originally Posted by MoMacFan

...
@motrek:
It was just a hypothesis - the external SSD I own and that is getting nicely warm is a 256Gig Thunderbolt drive with passive cooling (aluminum or magnesium body).
...

Ah, there's the culprit, I'm 99% sure all the heat is coming from the Thunderbolt controller chip. Not an issue for an internal drive obviously.

[dasx]

Originally Posted by MoMacFan

@dasx:
So it could be "normal"? At least the Mac Mini I have is no server model, so it might be normal and might explain why the official server Mini comes with a lower clock and without fusion drive...

Well. By "normal" I meant that it behaves as any other's Mini. Normal should've been Apple making it a little taller in order to put in a better cooling solution.

[philipma1957]

Originally Posted by dasx

Well. By "normal" I meant that it behaves as any other's Mini. Normal should've been Apple making it a little taller in order to put in a better cooling solution.

bingo the server is designed to fail from heat in 3 or more years if you push it and don't try to have extra cooling.

[dasx]

Originally Posted by philipma1957

bingo the server is designed to fail from heat in 3 or more years if you push it and don't try to have extra cooling.

Damn obsolescence!

[philipma1957]

Originally Posted by dasx

Damn obsolescence!

Thomas Edison made a lot of money selling 1000 hour light bulbs. But in his home the ones he had put in still work more then 50 years later. I will try to find a link.!!!

[dasx]

Originally Posted by philipma1957

Thomas Edison made a lot of money selling 1000 hour light bulbs. But in his home the ones he had put in still work more then 50 years later. I will try to find a link.!!!

Don't worry, I know the story. Man was clever, but I can still damn obsolescence. lol

[motrek]

Originally Posted by philipma1957

bingo the server is designed to fail from heat in 3 or more years if you push it and don't try to have extra cooling.

Do you have any data to back this up? I see plenty of old Apple hardware for sale on Craigslist that's still working fine, including Minis. I don't know if they were "pushed" or not but without any data to back up your claim I think it should be taken as pure speculation.

[propower]

Originally Posted by philipma1957

bingo the server is designed to fail from heat in 3 or more years if you push it and don't try to have extra cooling.

This is the one that interests me!

As a 25 year EE design vet I would agree that higher operating temps are more stressful. But the assertion that it will fail in "x" time does not to me appear to be a supportable conclusion.

Even if one did push a machine that hard day in day out is there any verifiable evidence of premature failure due to this? (The 2011 MBP line has similar operating temps). Have they been shown to fail in sharply disproportionate numbers?

Although Prime95 may very well be more stressful my 100% dev null test is miles beyond what I usually run. At what point is a synthetic benchmark just that and nothing else. Handbrake seems a better measure because at least this is a real world use. Even still one limit the threads used in handbrake to not just drive the machine at 100% across the board?

EDIT: Did dig around a little with 2011 macbook temp fail and there may be some evidence for increasing "logic board failures" that may be heat related. Would take a pile of reading to really ferret out. Hmmm...

[dasx]

Originally Posted by propower

Although Prime95 may very well be more stressful my 100% dev null test is miles beyond what I usually run. At what point is a synthetic benchmark just that and nothing else. Handbrake seems a better measure because at least this is a real world use. Even still one limit the threads used in handbrake to not just drive the machine at 100% across the board?

EDIT: Did dig around a little with 2011 macbook temp fail and there may be some evidence for increasing "logic board failures" that may be heat related. Would take a pile of reading to really ferret out. Hmmm...

As an engineer you'll then understand why using Prime. Just get to the worst scenario and work from there.

EDIT: Sorry, I don't know why I assumed you were an engineer from your post, lol
EDIT2: OK, EE. Is that it? Cause that's gotta be what made me think you were one too. Are you? In that case my post is valid.

[propower]

Originally Posted by dasx

As an engineer you'll then understand why using Prime. Just get to the worst scenario and work from there.

EDIT: Sorry, I don't know why I assumed you were an engineer from your post, lol
EDIT2: OK, EE. Is that it? Cause that's gotta be what made me think you were one too. Are you? In that case my post is valid.

:-)... yes I did high reliability 125degC power supply design for 25 years...
I am sure PRIME is a better test just not needed from my pov.

I am truly bothered by the design choice tradeoffs Apple is currently making in that they have great performing (and spec) machines that are beautiful to look at but have just enough cooling to keep them from overheating when driven above 25% load. From an old conservative designers pov this is just not great engineering. Unless they have really done their homework and thermally matched ALL the materials so that they can take the thermal expansion stresses. It can be done but is usually outside of the consumer field.

But I also allow that these may be just idealistic worries I have that may have no impact on real failure rate. Regardless I am very inclined to not keep any of these newer machines past the applecare expiration date...

[philipma1957]

Originally Posted by propower

:-)... yes I did high reliability 125degC power supply design for 25 years...
I am sure PRIME is a better test just not needed from my pov.

I am truly bothered by the design choice tradeoffs Apple is currently making in that they have great performing (and spec) machines that are beautiful to look at but have just enough cooling to keep them from overheating when driven above 25% load. From an old conservative designers pov this is just not great engineering. Unless they have really done their homework and thermally matched ALL the materials so that they can take the thermal expansion stresses. It can be done but is usually outside of the consumer field.

But I also allow that these may be just idealistic worries I have that may have no impact on real failure rate. Regardless I am very inclined to not keep any of these newer machines past the applecare expiration date...

I sell off my apple gear with time on the warranty. but the 2011 and the 2012 mini servers are really hot machines. my fear is mobo death on the quads.


note this is not a typical user that needs to worry .

this is a 24/7 user with cpu's pushed near and above 90%.

most mac rumor users have no need to worry.

[RiderX]

Thanks, Philipma!

I made a quick test with looping shells, 8 threads. So all cores were on 100% for about 15 minutes.

Result, the fan was above 4000, but never reached 5000 and the Cpu temp was round 95C.

This is way more for what I use my mini, and I thought I can live with that.

[MJL]

Originally Posted by dasx

Sure but it looks a little thick.
I honestly think that's gotta be blocking some air.

Not blocking any air at all - they use it for dirt bikes racing: this is specialised air filter foam, not the normal junk for chairs, pillows etc. The 2011 accumulates quite a bit of dust underneath the fan on electronic components affecting the heat dissipation. It is not targetted to keep the fan clean but to keep the dirt of the components. (cleaning the fan is easy but the cleaning the logic board on both sides is harder)

At idle I did not notice a difference on my 2.5 Ghz 2011 (mid) mac mini. However under load it made at 80C about 4 C difference (down to 76).

I did stick electronics heat sinks with adhesive thermal compound to the sides and that made a difference which means that the design is only geared to intermitted 100% CPU use and not to continued 100% CPU use what I do at times for backtesting (sometimes for days on end....)

One thing to be alert of is that different SSD's have large differences in power consumption.

If you drill holes in the bottom make sure to have the feet either on the corners of the aluminium housing or otherwise close to the three studs. PS The vent is located close to the HDD / WiFi and it does cool those parts as well.

In the end I grabbed a hardly used 2010 Mac mini server with two years warranty left and sold the mid 2011 Mac mini, best thing I ever did.

----------

Originally Posted by philipma1957

most mac rumor users have no need to worry.

agree unless they use handbrake a lot

[Snowcake]

Originally Posted by philipma1957

it half crashes. lets say you have the dual core running all '4' cores at 100% lets say 75% into the tests with 750 tests done and 250 tests left. it will freeze up and the last 25% or 250 tests don't get done. the '4' cores may show 100% or better yet one of the '4' cores will show 100% and will shift from the first core to the second core to the third core to the fourth core.

the last 25% of the tests do not get done. the machine cools done get a bit since only one of the '4' cores is stressed. I was able to get this to happen on both all 2012 minis using this program

http://www.macupdate.com/app/mac/23935/cputest

this did not happen when I tested the 2010 2.4 mac mini, but on the 2 quads and the one dual core 2012 it happened. I think it had trouble with the hyper threading or maybe the mountain lion osx.

the 2010 mini was using snow. the 2012 minis did this with the stock hdd the usb3 ssd and the t-bolt ssd. so for now I am not using cputest results on new minis. when I ran prime95 set on torture test small 1 2012 quad failed with rounding errors and 1 2012 quad did not get rounding errors the 1 2012 dual passed.

http://mersenne.org/freesoft/default.php

this is the link for prime95 set on torture test small

the 2012 mini does pass well one quad and one dual passes. I returned the quad that failed.

So if i understand it correctly, the strange behavior of cputest is the bug, not the actual "failed tests"?
In the picture above you see a section which displays "Failed tests"
The number of that was always 0 during the bug? It has never showed a another number than 0 on your second/replacement Quad Mac Mini? Correctly said, it had no errors but your problem was something else? (A bug)

[MJL]

Originally Posted by propower

This is the one that interests me!

As a 25 year EE design vet I would agree that higher operating temps are more stressful. But the assertion that it will fail in "x" time does not to me appear to be a supportable conclusion.

As an EE you do know that electrolytic capacitors have a rated life of a certain number of hours at a certain maximum temperature with a certain maximum ripple and that the life is very much affected by the operating temperature and ripple. The IBM T-series Thinkpads were "over engineered" and I know of some of these laptops still running 10 years later on original batteries..... (I ran a couple of them for 3 years almost daily at 100% and they are still running today 7 years later with the guy I sold them to).

[dasx]

Originally Posted by propower

I am truly bothered by the design choice tradeoffs Apple is currently making in that they have great performing (and spec) machines that are beautiful to look at but have just enough cooling to keep them from overheating when driven above 25% load. From an old conservative designers pov this is just not great engineering. Unless they have really done their homework and thermally matched ALL the materials so that they can take the thermal expansion stresses. It can be done but is usually outside of the consumer field.

But I also allow that these may be just idealistic worries I have that may have no impact on real failure rate. Regardless I am very inclined to not keep any of these newer machines past the applecare expiration date...

Yes. Mainly because with just 500um more they could've added a much better heatsink and a better fan. Not to mention 2 or 3mm longer and being able to put a larger fan. And tbh, no user would notice that at simple sight…

That's why I plan to mod it. I don't wanna be worried about temps. Also I don't wanna have a jet taking off under my desk every time I encode something using handbrake.

If I get to lower temps just 5ºC but manage to keep the fans in the lows 4500rpm instead of 5500rpm I'll be more than happy.

Originally Posted by MJL

If you drill holes in the bottom make sure to have the feet either on the corners of the aluminium housing or otherwise close to the three studs.

Why? I was planning to drill THIS zone (where the antenna plate is and where air is swallowed in) and put 3 rubber stands HERE. Is that a bad idea?

(See attached photo)

[MJL]

Originally Posted by dasx

Why? I was planning to drill THIS zone (where the antenna plate is and where air is swallowed in) and put 3 rubber stands HERE. Is that a bad idea?

(See attached photo)

Good locations - if you place feet on the area where there is no metal backing , i.e. somewhere in the location where the antenna grille is, then the bottom will flex.

Just be aware that cables attached to the back will want to tip the Mac mini - it is one thing that the airfilter solved (no feet required, it just rests on the filter).

On another note: if you are game enough you may want to change the thermal compound to Tuniq T4 - it is very good when the heatsink does not fit closely to the CPU / GPU. If I remember correctly it prevented the machine from hitting 100 C and it evened the temperature spikes considerably out. Some may suggest Arctic silver but that requires a very close fit between CPU and heatsink which was not the case with my computer (I normally use Arctic Silver but went out specifically for Tuniq T-4). It is also about the same consistency and color as the stuff Apple uses so it is not obvious that you've changed it....

I view the newer Mac mini's as a consumer machine, not as a business / workstation computer. (in other words no contineous high workload) however the low noise allows me to concentrate and it is easier to carry with me than a laptop when travelling (more robust - had several laptops damaged in overhead lockers on airplanes by inconsiderate travellers) The earlier ones were fine, not as much heat inside.

[dasx]

Originally Posted by MJL

Good locations - if you place feet on the area where there is no metal backing , i.e. somewhere in the location where the antenna grille is, then the bottom will flex.

Ahhh OK, Gotcha. So I should put the stands somewhere inside THIS perimeter, right?
(at the other side of course…)

[philipma1957]

Originally Posted by dasx

Ahhh OK, Gotcha. So I should put the stands somewhere inside THIS perimeter, right?
(at the other side of course…)

wait I have a link for a better way.



http://www.ebay.com/itm/D54-Brand-Ne...-/221112327607

see if you can find these in spain they are sturdy all metal. you can keep the stock piece in the box and drill lots of holes in this as it is better built. I own one and use it for a wall mount