Mini 2012 Base vs Mid - Thermal / noise performance

Beezer900 · Oct 31, 2012

Trying to decide between the base (dual core) or mid (quad core). Will be recording audio, so silent operation is a high priority.

I am leaning towards the mid (quad core), but I'm concerned that the increased power for quad core will result in higher temperatures and fan noise.

As I being overly cautious? Is it correct that the base (dual core) would be the more silent machine under normal use?

I currently use a 2006 core2duo macbook, and the fans get loud with pretty modest web use. Any kind of video will set them off. Can I expect the minis to be silent under anything other than extreme stress (80%+ CPU usage)?

tears2040 · Oct 31, 2012

You should first and foremost think about computer performance. The quad core Mac Mini will be faster than the base and a lot faster than your current setup.

Furthermore Mac Mini is basically silent and even if there is noise, you being a sound engineer I assume you should easily know how to kill low frequencies in your recording. Even then I HIGHLY doubt your mic will pic up any audio from a Mac Mini.

OldSchoolMacGuy · Oct 31, 2012

Think you're bring a bit overly cautious. Any mini is pretty much silent all the time. Either will be fine but I'd go the quad-core route. While it can consume more power requiring the fan speed to increase, it requires less of it's total power to complete the same tasks.

Think of a 100HP car and a 500HP car getting on the highway. Both have the same goal, get to 60mph, but the 500HP vehicle has to use less of it's total power to reach that speed. While the 100HP vehicle has to go through all the gears and rev things way up. High RPMs mean increased engine temp requiring the fan to kick on to cool things down. The higher HP car can hit highway speed in a single gear while keeping the RPMs well below 3000.

Poki · Oct 31, 2012

Even with full CPU load and 90°C temperature, the fan only goes to 3500 rpm, which shouldn't be noticeable with a good mic. And after all, the fans only turn up that much when you do heavy video editing. Everything else, including heavy photo editing, don't turn the fans up at all.

lilsoccakid74 · Oct 31, 2012

I think if you talk to a lot of sandy bridge mini owners, you find that the computer is far from silent when under stress. I believe this comes largely from the horrible thermal-paste application on them. With my 2.5 model w/ radeon6630, the CPU will fly up to 90*C+ while under 50%+ load. This can cause the fans to hit 5000 rpms, which sounds like a helicopter taking off

I believe Ivy Bridge will solve many of these heat-related issues. With the mini sitting 2 feet from me on my desk, I can only hear the fan once it hits 3000+rpms. I also found while using SMC fan control to raise my minimum rpms from 1800 to 2400, I can keep the mini silent, and the minis fan will "rev up (3k+ rpms)" much less often due to the increased airflow.

Hope this helped!

propower · Oct 31, 2012

I am an audio recording person also... I totally understand.

CPU load is the biggest thing. Go quad core absolutely.

I believe if CPU load steady state is kept in the 15 to 20% range the fans should stay at minimum - pretty close to silent. If one has to run higher CPU load a strategy I plan to employ is leaving the bottom cover off and making a stand for the mini with a big low rpm Silenx fan under it. Should be no more than 15dBA (very quiet) and solve the problem. I have seen others do a similar thing with remarkable temperature improvements.

I will be testing this very thing out next week

once my mini gets here (2.6/256SSD/16G Gskill Ram.)

pxbr · Nov 3, 2012

i'm no audio engineer - but, my new 2012 mac mini quad 2.3 is wayyy quieter under cpu load than my old 2011 mac mini (using handbrake which is a really easy way to stress the cpu!)

Beezer900 · Nov 3, 2012

I'm no audio engineer either, but even for amateur home recordings, fan noise is enough to really spoil things.

Thanks for the replies everyone. The quad core sounds like a good choice and I feel quite reassured now!

dasx · Nov 3, 2012

Using handbrake I got my 2.6 i7 to 105ºC (221ºF). Pretty damn hot. Fans kicked in at almost 4000rpm and they were "kind of" loud. Not annoying or disturbing but you could clearly hear them. I guess I wouldn't mind if this thing wasn't super quiet when idle.

So yes, you do notice them kicking in under FULL load. It's not something spectacular, it's just that you notice it as it's usually silent.

Will it make recordings bad? Well I can't tell. I don't think the CPU should get that hot during recordings, but who am I to know.

Beezer900 · Nov 3, 2012

I won't be getting anywhere near full CPU load, so it should be fine.

Out of interest, roughly what % CPU usage do the fans kick in at an audible level?

hugodrax · Nov 3, 2012

Heavy use and its silent (i7 2.6) So silent that now I notice the fan noise on my Smart UPS 1500LCD and also the sound of my External drive

dasx · Nov 3, 2012

Beezer900 said:
I won't be getting anywhere near full CPU load, so it should be fine.

Out of interest, roughly what % CPU usage do the fans kick in at an audible level?

Doens't really need any CPU usage I think. I was doing some conversions with subler some minutes ago and noticed that the fans kicked in while the CPU was at like 30%. Checked temps and they were at around 95ºC.

It seems that all it takes to speed up the fans is one CPU to get hot. For example:
Imagine you run app X which puts one core to 100% and leaves all the others to 0%. That means an overall CPU load of 12.5%. I think that if you stay that way, the cores which is at 100% will eventually get the whole chip to 95ºC or so with its own heat dissipation. Therefore fans will kick in, eventually.

^^^^That's all guess. Maybe it needs at least 2/3 cores to make the whole chip get hot, but I made my point.

----------

hugodrax said:
Heavy use and its silent (i7 2.6) So silent that now I notice the fan noise on my Smart UPS 1500LCD and also the sound of my External drive

Get a 1080p video file in MKV with DTS audio and convert it to a high profile handbrake preset. You'll hear it trying to cool himself down.

Maxx Power · Nov 12, 2012

dasx said:
Doens't really need any CPU usage I think. I was doing some conversions with subler some minutes ago and noticed that the fans kicked in while the CPU was at like 30%. Checked temps and they were at around 95ºC.

It seems that all it takes to speed up the fans is one CPU to get hot. For example:
Imagine you run app X which puts one core to 100% and leaves all the others to 0%. That means an overall CPU load of 12.5%. I think that if you stay that way, the cores which is at 100% will eventually get the whole chip to 95ºC or so with its own heat dissipation. Therefore fans will kick in, eventually.

^^^^That's all guess. Maybe it needs at least 2/3 cores to make the whole chip get hot, but I made my point.

That's not how it works for Intel chips with dynamic TDP allocation, a.k.a., Turbo.

If only 1 core is 100% taxed, the other cores' worth of thermal-dissipation and power consumption headroom up to the max TDP of the chip is re-directed dynamically by Turboing the working core a few bins (multiples of the base clock - 100mhz) higher. The TDP of the chips in the quad-core minis are 45 watts, while the dual core base model is only 35 watts.

So whether or not you only load 1 core to 100% or 2 cores to 100%, the thermal dissipation power of the chip under either circumstance is 35 watts. The quad-core variants have a much more aggressive 1-core and 2-core active Turbo because there is more TDP room to begin with. Although in reality, the total power dissipated with only 1 core on Turbo is likely lower than 2 core at 100%, but Intel's algorithms for Turbo and design guides for heatsink/fan assemblies are calculated and based on TDP.

This means, and according to the Apple specifications, the quad-core minis run typically louder. Since they use the same heatsink/fan assembly, this reasons to that the fan speed curves/minimum fan speeds are likely higher in the quad-core models to dissipate the additional 10 watts of TDP. See this snippet:

"Typical acoustical performance, sound pressure level (operator position):

Dual-core model: 12 dBA at idle
Quad-core model: 15 dBA at idle
"

propower · Nov 12, 2012

There are no silent computers

There are very few silent rooms!

Typical background noise in a quiet room is ~30dBA. FWIW my rooms measure ~32dBA. At idle I can't hear my mini above the background. To keep it at idle (lowest fan) I have to keep my sessions under 18% total cpu load. There are many strategies to do this. Mix as you go and print tracks. Make guide tracks so that resources are kept low etc. At 25% load this creeps up to ~42dBA. Yes I can pick out that sound. But I can also mix or with careful mic placement record pretty much any acoustic instrument.

So silence is golden but very hard to come by. If the above is not enough the mini has yet another easy solution! Since it is relatively so small, build a well ventilated doghouse to put the mini in! Acoustically treat the inside and use enough low rpm quiet fans form Silenx or others.

dasx · Nov 12, 2012

Maxx Power said:
That's not how it works for Intel chips with dynamic TDP allocation, a.k.a., Turbo.

If only 1 core is 100% taxed, the other cores' worth of thermal-dissipation and power consumption headroom up to the max TDP of the chip is re-directed dynamically by Turboing the working core a few bins (multiples of the base clock - 100mhz) higher. The TDP of the chips in the quad-core minis are 45 watts, while the dual core base model is only 35 watts.

So whether or not you only load 1 core to 100% or 2 cores to 100%, the thermal dissipation power of the chip under either circumstance is 35 watts. The quad-core variants have a much more aggressive 1-core and 2-core active Turbo because there is more TDP room to begin with. Although in reality, the total power dissipated with only 1 core on Turbo is likely lower than 2 core at 100%, but Intel's algorithms for Turbo and design guides for heatsink/fan assemblies are calculated and based on TDP.

This means, and according to the Apple specifications, the quad-core minis run typically louder. Since they use the same heatsink/fan assembly, this reasons to that the fan speed curves/minimum fan speeds are likely higher in the quad-core models to dissipate the additional 10 watts of TDP. See this snippet:

"Typical acoustical performance, sound pressure level (operator position):

Dual-core model: 12 dBA at idle
Quad-core model: 15 dBA at idle
"

Nice explanation. Thanks!

Maxx Power · Nov 12, 2012

dasx said:
Nice explanation. Thanks!

No problem!

Kaloman · Nov 13, 2012

lilsoccakid74 said:
I think if you talk to a lot of sandy bridge mini owners, you find that the computer is far from silent when under stress. I believe this comes largely from the horrible thermal-paste application on them. With my 2.5 model w/ radeon6630, the CPU will fly up to 90*C+ while under 50%+ load. This can cause the fans to hit 5000 rpms, which sounds like a helicopter taking off

I believe Ivy Bridge will solve many of these heat-related issues. With the mini sitting 2 feet from me on my desk, I can only hear the fan once it hits 3000+rpms. I also found while using SMC fan control to raise my minimum rpms from 1800 to 2400, I can keep the mini silent, and the minis fan will "rev up (3k+ rpms)" much less often due to the increased airflow.

Hope this helped!

Why Does Ivy Bridge Run Hotter?
Why should the Ivy Bridge chips, which are more power efficient than Sandy Bridge, run hotter?

PCWorld Senior Editor Jason Cross suggests four possible reasons for these findings. The heat issue also may be a combination of these factors:

Ivy Bridge packs more transistors into a smaller area than Sandy Bridge. That means increased thermal density--heat generated in a given area. It's harder to cool processors where the heat is concentrated into a small space.

Modern CPU chips have a "heat spreader" on top of them. It's part of the processor package. The Sandy Bridge heat spreader was soldered on, while the Ivy Bridge heat spreader makes contact with the CPU using thermal paste. It may not be conducting heat as well as the Sandy Bridge's soldered heat spreader.

Ivy Bridge usually runs using a lower operating voltage than Sandy Bridge. But to overclock it way up to 4.9GHz, you have to bump up the voltage. In fact you have to add more voltage to an Ivy Bridge processor than a Sandy Bridge processor to reach that level. Since power consumption is strongly related to operating voltage, this means you're adding more power consumption and heat.

Intel's 22nm manufacturing process is quite new, while the 32nm manufacturing process used with Sandy Bridge is more mature. As the months roll on and Intel improves its 22nm process, leakage and heat dissipation may improve.

http://www.pcworld.com/article/2547...ps_appear_to_run_hotter_when_overclocked.html

motrek · Nov 13, 2012

Beezer900 said:
...
I currently use a 2006 core2duo macbook, and the fans get loud with pretty modest web use. Any kind of video will set them off. Can I expect the minis to be silent under anything other than extreme stress (80%+ CPU usage)?

Unfortunately if you check Activity Monitor you will find that watching most Flash-based video on web pages is hardly "modest" use and will usually cause one or even two cores to be utilized 100%.

Flash is really terrible when it comes to using the CPU. Watching a video with QuickTime or VLC will hardly cause a noticeable blip of CPU usage, but watching a YouTube video seems to take more processing power than curing cancer.

There are many utilities like "FlashFrozen" that automatically kills Flash if you don't want to use it, so random ads on web pages don't take up all your battery power and/or make your machine sound like a vacuum cleaner.

Liquidstate · Nov 13, 2012

Definitely a topic I was waiting for...

I have to move forward with my maxed out audio recording system. I've been juggling needs and formats (I have legacy gear). Who knows what a new Mac Pro would be, or when, or what Avid/ProTools is going to do next? I have stacks of system design diagrams.

Back on topic, the noise level of the Mac is important to me. That's why I moved to an iMac, and taught myself the PT workarounds. It's quiet enough, but it's underpowered and too old to upgrade.

I've been seriously considering a Mac Mini 2.6 i7 16GB. But I have some concerns it might be a little too noisy.

I'm starting to wonder if the slightly slower Mac Mini 2.3 i7 quad SSD with 16 GB might run a little cooler and quieter?

Is there any technical justification for this idea?

Maxx Power · Nov 13, 2012

Kaloman said:
Why Does Ivy Bridge Run Hotter?
Why should the Ivy Bridge chips, which are more power efficient than Sandy Bridge, run hotter?

PCWorld Senior Editor Jason Cross suggests four possible reasons for these findings. The heat issue also may be a combination of these factors:

Ivy Bridge packs more transistors into a smaller area than Sandy Bridge. That means increased thermal density--heat generated in a given area. It's harder to cool processors where the heat is concentrated into a small space.

Modern CPU chips have a "heat spreader" on top of them. It's part of the processor package. The Sandy Bridge heat spreader was soldered on, while the Ivy Bridge heat spreader makes contact with the CPU using thermal paste. It may not be conducting heat as well as the Sandy Bridge's soldered heat spreader.

Ivy Bridge usually runs using a lower operating voltage than Sandy Bridge. But to overclock it way up to 4.9GHz, you have to bump up the voltage. In fact you have to add more voltage to an Ivy Bridge processor than a Sandy Bridge processor to reach that level. Since power consumption is strongly related to operating voltage, this means you're adding more power consumption and heat.

Intel's 22nm manufacturing process is quite new, while the 32nm manufacturing process used with Sandy Bridge is more mature. As the months roll on and Intel improves its 22nm process, leakage and heat dissipation may improve.

http://www.pcworld.com/article/2547...ps_appear_to_run_hotter_when_overclocked.html

The heatspreader-bonding theory has had substantial amount of test results and it indeed affects desktop grade CPUs (that have heatspreaders).

Mobile Ivy Bridge as used in MBP, Mac Minis and eventually iMacs do NOT have integrated heatspreaders. They are naked-dies only.

As such the mobile Ivy Bridge CPUs should not run substantially hotter due to this.

motrek · Nov 13, 2012

counterbalance said:
I have to move forward with my maxed out audio recording system. I've been juggling needs and formats (I have legacy gear). Who knows what a new Mac Pro would be, or when, or what Avid/ProTools is going to do next? I have stacks of system design diagrams.

Back on topic, the noise level of the Mac is important to me. That's why I moved to an iMac, and taught myself the PT workarounds. It's quiet enough, but it's underpowered and too old to upgrade.

I've been seriously considering a Mac Mini 2.6 i7 16GB. But I have some concerns it might be a little too noisy.

I'm starting to wonder if the slightly slower Mac Mini 2.3 i7 quad SSD with 16 GB might run a little cooler and quieter?

Is there any technical justification for this idea?

Interesting question. You can double check me easily enough but I believe both chips have the same TDP at 45W so according to Intel's specifications they should be exactly as loud as each other under load.

Since you would expect a faster chip to run hotter, there could be a few reasons for this odd specification that I can think of:

1) Maybe the 2.6GHz chips are bin sorted so they can run correctly at a lower voltage than the 2.3GHz chips, so they both really do use the same power (and thus make the same amount of noise).

2) Maybe the 2.3GHz chip does use somewhat less than 45W but Intel wanted to round up just to make things simpler for engineers designing around these chips.

3) Maybe one or both chips reduce their performance (turbo boost) to stay in the 45W envelope, meaning the 2.6GHz chip might run at the same speed as the 2.3GHz chip in more cases than one might think.

Although ultimately I wonder what the Mini will be doing while recording? Just grabbing sound data and storing it somewhere should not be taxing to the CPU and thus the Mini should basically be silent. Mixing all the tracks might stress the CPU but is that done at the same time as recording?

propower · Nov 13, 2012

counterbalance said:
I've been seriously considering a Mac Mini 2.6 i7 16GB. But I have some concerns it might be a little too noisy.

I'm starting to wonder if the slightly slower Mac Mini 2.3 i7 quad SSD with 16 GB might run a little cooler and quieter?

Is there any technical justification for this idea?

I have posted in several places (including above) test results for noise from the 2.6 i7 mini using pro tools (Just search for my other posts).

In my opinion the difference between the 2.3 and 2.6 will be very subtle for this use. There may be a slight edge for the 2.3 though. Both procs will have standby heat generation (just by sitting there) and I would expect the 2.3 to be slightly less. Once you are using pro tools there is a finite number of things the proc must do, be it at 2.3 or 2.6 so I would guess the heat generation for this part will be the same!

After seeing the loads I get on the 2.6 I am happy with my choice. The noise is within what I expected and I have many avenues to reduce it in practice (dog house for the mini, alternate heat sinking, fans, cold plates, removing the bottom cover, orientation, getting it 6 feet away instead of 2 etc etc....

14 days to try one.... you could always just give it a whirl !

Maxx Power · Nov 13, 2012

motrek said:
Interesting question. You can double check me easily enough but I believe both chips have the same TDP at 45W so according to Intel's specifications they should be exactly as loud as each other under load.

Since you would expect a faster chip to run hotter, there could be a few reasons for this odd specification that I can think of:

1) Maybe the 2.6GHz chips are bin sorted so they can run correctly at a lower voltage than the 2.3GHz chips, so they both really do use the same power (and thus make the same amount of noise).

2) Maybe the 2.3GHz chip does use somewhat less than 45W but Intel wanted to round up just to make things simpler for engineers designing around these chips.

3) Maybe one or both chips reduce their performance (turbo boost) to stay in the 45W envelope, meaning the 2.6GHz chip might run at the same speed as the 2.3GHz chip in more cases than one might think.

Although ultimately I wonder what the Mini will be doing while recording? Just grabbing sound data and storing it somewhere should not be taxing to the CPU and thus the Mini should basically be silent. Mixing all the tracks might stress the CPU but is that done at the same time as recording?

As far as why 2 different speed chips have the same TDP rating goes, the reason is that the TDP number (so say, 45 Watts), is a BALL-PARK figure that Intel designates so that engineers who are designing systems and cooling will know how much they must dissipate to operate the chip to its specifications (to not throttle ALL the time).

In reality, the 2.3 Ghz chip have a SLIGHTLY lower TDP than the 2.6 Ghz chip if the silicon they came from are identical, and the voltages they operate at are identical. Although this isn't always guaranteed since Intel can select and bin chips according to their leakage and frequency headroom, no two silicon dies are exactly the same... In practice, what Intel's chips do is that they internally monitor their current power consumption and temperatures to predict its own current TDP and uses an algorithm to throttle the CPU speeds when the temperatures gets too high so that in essence, the CPU is dissipating more TDP than the heatsink/fan is capable of. This is what saves Intel silicon's bacon in MOST consumer laptops. You'll notice that the Intel CPUs runs very HOT, from Dell, to HP to Apple. It doesn't get destroyed because it can throttle very quickly to prevent thermal catastrophe. As per Anandtech's tests, the MBP Retina (both 13 and 15 inches) are some of the ONLY laptops with performance oriented Intel mobile chips that don't throttle (in the case of the 13 inch) or throttle lightly (in the case of the 15 inch), so the the performance of the system when loaded remains constant. On some systems (the new Dell XPS laptops comes to mind as per Anandtech), the throttle is horrible that you practically only get what Intel intended for performance (Turbo) a tiny fraction of the time. This has been the norm since, I think, Sandy Bridge. Last time I read about this, the Mac Mini CPUs throttle too, so you'll notice a reduction in Turbo as the minutes drags on into a CPU intensive workload as the CPU gets hot. Previously, there were some workarounds in software to artificially hold the Turbo, but I'm not convinced that this action is healthy for the CPU.

OCCASIONALLY, Intel or AMD have been known to produce different VID (Vcore ID) chips, so that 2 chips are both rated at 45Watts, for example, but one chip has a Vcore at 1.2Volts while the other has only 1.15Volts. In this case, the lower VID variant runs cooler. However, I'm not aware that Intel is doing this with their modern Ivy Bridge chips...

dasx · Nov 13, 2012

I got a 2.6GHz i7 and it does get loud… when using 100% of the CPU. Does the software you use consume a lot of CPU? In that case the DC i5 might be a better choice. But if not using 100% of CPU, then the i7 would be fine.

I don't know what software you use, but if it's prepared to work multithread and it does use all CPU power when available… Then expect the Mini to be damn loud. And I mean LOUD. Fans at 5500rpm are sooo noisy.

Gotta say though, that I only achieved that using Prime95, a software designed to stress the CPU to its maximum. When running some programs I make which also consume 100% of CPU (but not as intense as Prime95, they just use 8 threads with simple tasks but taking all the power) temps stayed like in Prime95 (around 98ºC) but fan speeds where much lower, around 3500rpm. Still hearable, but not like at 5500rpm.

Also, when using CPUTest (another CPU stresser), after 20 minutes with the CPU at 100% it was at 97-100ºC and fans stayed at 4000rpm. Much silent than with Prime.

Snowcake · Nov 13, 2012

Maxx Power said:
As far as why 2 different speed chips have the same TDP rating goes, the reason is that the TDP number (so say, 45 Watts), is a BALL-PARK figure that Intel designates so that engineers who are designing systems and cooling will know how much they must dissipate to operate the chip to its specifications (to not throttle ALL the time).

In reality, the 2.3 Ghz chip have a SLIGHTLY lower TDP than the 2.6 Ghz chip if the silicon they came from are identical, and the voltages they operate at are identical. Although this isn't always guaranteed since Intel can select and bin chips according to their leakage and frequency headroom, no two silicon dies are exactly the same... In practice, what Intel's chips do is that they internally monitor their current power consumption and temperatures to predict its own current TDP and uses an algorithm to throttle the CPU speeds when the temperatures gets too high so that in essence, the CPU is dissipating more TDP than the heatsink/fan is capable of. This is what saves Intel silicon's bacon in MOST consumer laptops. You'll notice that the Intel CPUs runs very HOT, from Dell, to HP to Apple. It doesn't get destroyed because it can throttle very quickly to prevent thermal catastrophe. As per Anandtech's tests, the MBP Retina (both 13 and 15 inches) are some of the ONLY laptops with performance oriented Intel mobile chips that don't throttle (in the case of the 13 inch) or throttle lightly (in the case of the 15 inch), so the the performance of the system when loaded remains constant. On some systems (the new Dell XPS laptops comes to mind as per Anandtech), the throttle is horrible that you practically only get what Intel intended for performance (Turbo) a tiny fraction of the time. This has been the norm since, I think, Sandy Bridge. Last time I read about this, the Mac Mini CPUs throttle too, so you'll notice a reduction in Turbo as the minutes drags on into a CPU intensive workload as the CPU gets hot. Previously, there were some workarounds in software to artificially hold the Turbo, but I'm not convinced that this action is healthy for the CPU.

OCCASIONALLY, Intel or AMD have been known to produce different VID (Vcore ID) chips, so that 2 chips are both rated at 45Watts, for example, but one chip has a Vcore at 1.2Volts while the other has only 1.15Volts. In this case, the lower VID variant runs cooler. However, I'm not aware that Intel is doing this with their modern Ivy Bridge chips...

The cpu not using Turboboost is not throttling.
Throttling is: Working at a lower speed than the standard speed. TurboBoost is above standard speed.

Mini 2012 Base vs Mid - Thermal / noise performance

macrumors newbie

macrumors 6502

Suspended

macrumors 65816

macrumors 6502

macrumors 6502a

macrumors member

macrumors newbie

macrumors 65816

macrumors newbie

macrumors 65816

macrumors 65816

Cancelled

macrumors 6502a

macrumors 65816

Cancelled

macrumors newbie

macrumors 68030

macrumors member

Cancelled

macrumors 68030

macrumors 6502a

Cancelled

macrumors 65816

macrumors regular

Our Staff