Apple Engineers Explain New Mac Pro's Innovative Cooling Features

[TracesOfArsenic]

Of course it would also open up for the sale of magical Apple coolant, probably a missed opportunity 🌊

There are enough people already drinking the Apple kool-aid, let's not give them a literal product.

 

[cmaier]

most high end premium air cooler like noctua and cryorig have very high fin density, the fin density on the mac pro looks like these cheap 20 dollar cooler master air coolers.

Just googled to refresh my recollection. See, e.g. https://web.mit.edu/16.unified/www/FALL/thermodynamics/notes/node128.html

”fin density” doesn’t really enter into it. What matters (assuming identical materials) is fin height, number of fins, and the rate of flow of whatever fluid we’re talking about (air, in this case).

You are also dissipating heat that is generated from some surface area, so for a given surface you need to calculate how much heat is generated in that area, then determine the number of fins, the flow rate, and the height of the fins. You can vary each independently to pick a solution you like. So, for example, apple may have a higher flow rate than these other solutions, or the heat-generating components may be spread out more so that less heat (and less fins) are needed in a given area, or they may have fins that are longer.

You really can’t tell much just by counting fins in a given area.

When I was designing electronics packages we used to figure we could air cool if the dissipation was less than 10W/square cm. (Of course that takes into account that you could use a heat spreader to get the power dissipation density under that number). Over that number and you had to start getting creative. I remember DEC had a “chimney” containing an aerosol that would heat up and rise, carrying away heat in a sealed “chimney,” then condense at the cool top and fall down again, creating a tiny little circulation, and it was able to handle some crazy power density.

 

[-hh]

The article states: "According to Apple, the reworked "cheese grater" look achieves around 20 percent more airflow compared to the Power Mac G5 that preceded it"

I'm confused, or am I reading that wrong? The cheese grater style system that directly preceded this was an intel xeon system, released in 2012. Not the G5.

Doesn't really matter, since the G5's and Xeon both used the same "cheese grater" hole pattern.

But in any event, this "+20%" claim is a YAWN because the engineering details are so darn obvious: when they changed the hole pattern, they also changed the ratio of {hole diameter} -vs- {web}

(FYI "web" is the amount of metal remaining between holes).

Eyeballing it, I'd say that the web on the cMP was roughly "r" (1 radius); whereas the new case looks to be around "r/3". In simple terms, there's more hole and less metal, thus more unitized cross-sectional area to facilitate airflow...not rocket science at all.

 

[-hh]

that fin density is weak looking,
View attachment 882359

Grabbing your pic for addressing the liquid cooling subthreads (en masse).

First, that pic clearly illustrates a thermal management system which is working as a heat spreader from the heat source on the left, and then out into fins to move the thermal load into a (presumably fan-based moving) air mass to get the heat overboard.

Two things on this.

First, those four heat pipes are an integral part of the thermal transport mechanism. I've not looked yet into the technical weeds, but odds are pretty good that they're liquid-filled in some form, even though no pump is evident: they could be capillary, etc.

In the past, I've noted that the no-liquids (and zero moving parts) technology of TPG could serve in this role for transporting heat load sources of up to 200W/cm^2 out into a fin assembly like this with zero moving parts.

Conclusion: until one gets cut open, we can't conclude "no liquids".

Second, the presence of a heat spreader attached onto whatever-that-thing-is-on-the-left does have a material impact on all discussions of ease-of-upgradability. In a nutshell, it doesn't make it easier.

Note:
TPG = Thermal Pyrolytic Graphite. Its a solid which has heat transfer coefficients nearly 4x better than copper (~1500 W/m versus ~400 W/m·K); the catch is that its not symmetrical in all three dimensions, but only in 2 of its 3 dimensions, which makes it harder to design a spreader.

 

[dogslobber]

Typical NIH syndrome from Apple. You know there will be a multitude of recalls due to the fans blowing hot air to the wrong parts. It's all hot air since if you can't get a keyboard to work correctly across 3 generations then how would you know how to cool something down?

 

[UK-MacAddict]

The article states: "According to Apple, the reworked "cheese grater" look achieves around 20 percent more airflow compared to the Power Mac G5 that preceded it"

I'm confused, or am I reading that wrong? The cheese grater style system that directly preceded this was an intel xeon system, released in 2012. Not the G5.

The original cheesegrater design was released with the Power Mac G5 when Apple were still running Power PC CPU's. Even after switching to Intel they kept the same chassis so I think the cooling system remained the same.

 

[IG88]

And yes, those Xeon's... which is silly because they're becoming less and less competitive lately.

Agreed for sure. The Xeon tax appears worse than the Apple tax. Too bad there isn't a prosumer version of this box in a AMD 3950X or Threadripper. Would require a different chipset altogether though.

I still think there's a market for a non-iMac Mac tower. Apple may not agree.

 

[Justanotherfanboy]

Both?

View attachment 882419

Hahahaha!
Excellent. Ok, kudos.... you actually revived a joke I thought couldn’t get me to crack a smile again. =)

 

[sakurarain]

I am an obsessive-compulsive “perfectionist”, I am concerning how to clean the ventilation holes and how long it takes to make a through cleaning.......

 

[Abazigal]

Typical NIH syndrome from Apple. You know there will be a multitude of recalls due to the fans blowing hot air to the wrong parts. It's all hot air since if you can't get a keyboard to work correctly across 3 generations then how would you know how to cool something down?

Since it’s all false equivalencies either way...

Not having a working keyboard didn’t stop Apple from making the Apple Watch and AirPods as successful as they were.

 

[FairlyKors]

that fin density is weak looking,
View attachment 882359

comparing to

I’m pretty sure Apple knows exactly what it’s doing regarding the fin density you’ve pointed out...

 

[ilikewhey]

I’m pretty sure Apple knows exactly what it’s doing regarding the fin density you’ve pointed out...

thats a strong statement consider apple has a pretty bad track records with cooling, mbp i9 overheating quickly comes to mind if you wanna argue if apple knows exactly what they are doing in the past.

 

[Abazigal]

problem is the market doesn't give a **** about the profit margins apple wants.

if Apple can't be competitive at that price and performance point, they may (i recognise I am speculating) have issues moving units at these price points.

I don't think there will be a problem with the upper tiers. Those who need those, know what they're getting and know why they're getting it. but it's these lower tiers, where Apple's over-designing is costing almost 2.5 times the price point for performance.

again: I am not judging the computer's specs. Nor it's hardware prowress. I think this is finally the Mac Pro that most of the Mac Pro users are absolutely looking for.

Just doing my typical Accounting cost to benefit analysis. And at the lower tiers, especially the "desktop" class performance levels, This is an over-engineered, and therefore comes with a pricetag that far exceeds the performance levels at this price point.

People can try and make excuses for it. Some valid. Some vapid.

Example of a vapid excuse: "Mac OS is worth the $2,000 premium". "they're just covering overhead costs of designing it"

If your design costs are so ridiculously high to essentially release a parity product to existing standards, than you have fundamental design and direction problems in your corporations direction.

So people elect not to get a cheaper Mac Pro option and instead go with a cheaper alternative such as an iMac Pro or iMac.

Seems like the problem has pretty much worked itself out?

I think you need to stop evaluating these products in a vacuum and instead pay more attention to where they stand in the entire lineup. Apple sells an entire lineup of products. Fewer people buying a certain product may not necessarily be a bad thing in itself if it means that this small clientele is opting for the higher-spec options (ie: more margins for Apple), while the rest go with more affordable Mac alternatives. At the end of the day, they remain in the ecosystem.

 

[Hazmat401]

I personally think the worry about dust is being blown up for no reason

Push come to shove.... hook up the crevice attachment to your vacuum cleaner hose and go crazy every other month

 

[ilikewhey]

Just googled to refresh my recollection. See, e.g. https://web.mit.edu/16.unified/www/FALL/thermodynamics/notes/node128.html

”fin density” doesn’t really enter into it. What matters (assuming identical materials) is fin height, number of fins, and the rate of flow of whatever fluid we’re talking about (air, in this case).

You are also dissipating heat that is generated from some surface area, so for a given surface you need to calculate how much heat is generated in that area, then determine the number of fins, the flow rate, and the height of the fins. You can vary each independently to pick a solution you like. So, for example, apple may have a higher flow rate than these other solutions, or the heat-generating components may be spread out more so that less heat (and less fins) are needed in a given area, or they may have fins that are longer.

You really can’t tell much just by counting fins in a given area.

When I was designing electronics packages we used to figure we could air cool if the dissipation was less than 10W/square cm. (Of course that takes into account that you could use a heat spreader to get the power dissipation density under that number). Over that number and you had to start getting creative. I remember DEC had a “chimney” containing an aerosol that would heat up and rise, carrying away heat in a sealed “chimney,” then condense at the cool top and fall down again, creating a tiny little circulation, and it was able to handle some crazy power density.

good point, its hard to determine just from first glance and without testing,

i'm confident apple is using the so called "closed chimney" heatpipes, since that has been the industry standard in pc cooling solution since ever, my noctua d14 has it as well as my old prolimatech megahlem from 2009. i'll be very surrpsied if apple doesn't use them.

now that we established everyone including apple uses this "closed chimney" solution. what is setting all the competitors apart are the fin density and materials used, copper being the best performer but heavy and more expensive. typically oems states copper if it has any since thats a good marketing ploy they like to use. i don't think apple is using copper heat pipes here.

Grabbing your pic for addressing the liquid cooling subthreads (en masse).

First, that pic clearly illustrates a thermal management system which is working as a heat spreader from the heat source on the left, and then out into fins to move the thermal load into a (presumably fan-based moving) air mass to get the heat overboard.

Two things on this.

First, those four heat pipes are an integral part of the thermal transport mechanism. I've not looked yet into the technical weeds, but odds are pretty good that they're liquid-filled in some form, even though no pump is evident: they could be capillary, etc.

In the past, I've noted that the no-liquids (and zero moving parts) technology of TPG could serve in this role for transporting heat load sources of up to 200W/cm^2 out into a fin assembly like this with zero moving parts.

Conclusion: until one gets cut open, we can't conclude "no liquids".

Second, the presence of a heat spreader attached onto whatever-that-thing-is-on-the-left does have a material impact on all discussions of ease-of-upgradability. In a nutshell, it doesn't make it easier.

Note:
TPG = Thermal Pyrolytic Graphite. Its a solid which has heat transfer coefficients nearly 4x better than copper (~1500 W/m versus ~400 W/m·K); the catch is that its not symmetrical in all three dimensions, but only in 2 of its 3 dimensions, which makes it harder to design a spreader.

you don't have to cut them open, they are filled with liquids, nvidia's vapor chamber used since gtx 1080 is a good illustration.

 

[Wash08]

I remember buying the first cheese grater Mac Pro years ago and I paid like $2400. They started at $1999 if I recall correctly. Why are they $6,000 now? Are they really that much better than a comparable Windows PC?

Beacuse of this

 

[cmaier]

good point, its hard to determine just from first glance and without testing,

i'm confident apple is using the so called "closed chimney" heatpipes, since that has been the industry standard in pc cooling solution since ever, my noctua d14 has it as well as my old prolimatech megahlem from 2009. i'll be very surrpsied if apple doesn't use them.

now that we established everyone including apple uses this "closed chimney" solution. what is setting all the competitors apart are the fin density and materials used, copper being the best performer but heavy and more expensive. typically oems states copper if it has any since thats a good marketing ploy they like to use. i don't think apple is using copper heat pipes here.


you don't have to cut them open, they are filled with liquids, nvidia's vapor chamber used since gtx 1080 is a good illustration.

no, there are no heat pipes. They’d be quite visible and Apple would have talked about them.

and for what it’s worth, diamond is better for heat sinks than copper, so quit buying those cheap PC copper heat sinks.

oh, and as indicated on those webpages which list the equations, density is not, per se, determinative. Given identical air flow rates, materials, and fin dimensions,then maybe it would be relevant (of course more density means less spacing which means slower air flow, which means you need a faster fan, so nothing is free.

 

[IG88]

I’m pretty sure Apple knows exactly what it’s doing regarding the fin density you’ve pointed out...

By that logic, Apple shouldn't have needed to redesign the cooling on the 16" MBP, that's using the exact same CPUs the predecessor had. Since Apple knows exactly what it's doing regarding cooling?

 

[ilikewhey]

no, there are no heat pipes. They’d be quite visible and Apple would have talked about them.

and for what it’s worth, diamond is better for heat sinks than copper, so quit buying those cheap PC copper heat sinks.

oh, and as indicated on those webpages which list the equations, density is not, per se, determinative. Given identical air flow rates, materials, and fin dimensions,then maybe it would be relevant (of course more density means less spacing which means slower air flow, which means you need a faster fan, so nothing is free.

no, not faster fan, more static pressure, meaning more optimized fan blades for pushing more air,

heatpipes, heatsink pipes, thermal exchanger, whatever you wanna call it.

 

[IG88]

that fin density is weak looking,

No fan direct mounted on the heat sink. My guess is they had to space the fins that far apart to get enough airflow through them. Time will tell if it's adequate.
[automerge]1576207130[/automerge]

most high end premium air cooler like noctua and cryorig have very high fin density, the fin density on the mac pro looks like these cheap 20 dollar cooler master air coolers.

Remember that Apple doesn't have a fan mounted on their heatsink, correct?

 

[cmaier]

no, not faster fan, more static pressure, meaning more optimized fan blades for pushing more air,

heatpipes, heatsink pipes, thermal exchanger, whatever you wanna call it.

True, doesn’t have to be faster as long as it increases air pressure, but it also has to be turbulent which affects the required speed.
And there is no sealed heat pipe, heat sink pipE, or whatever else you want to call it.

and carbon nanotubes are better than diamond, so stop buying those cheap PC diamond-finned radiators.

 

[ilikewhey]

No fan direct mounted on the heat sink. My guess is they had to space the fins that far apart to get enough airflow through them. Time will tell if it's adequate.
[automerge]1576207130[/automerge]

Remember that Apple doesn't have a fan mounted on their heatsink, correct?

my guess is given where they placed the cpu heatsink right infront of the chassis. so while the fans are not directly mounted on the heatsink, its close enough proximity should warrant justice for proper static air pressure.
[automerge]1576208407[/automerge]

True, doesn’t have to be faster as long as it increases air pressure, but it also has to be turbulent which affects the required speed.
And there is no sealed heat pipe, heat sink pipE, or whatever else you want to call it.

and carbon nanotubes are better than diamond, so stop buying those cheap PC diamond-finned radiators.

manufacturers are too cheap to even use copper, let along using sci fi lab testing material for mass producing electronic hardware 😂

 

[IG88]

no, there are no heat pipes. They’d be quite visible and Apple would have talked about them.

No heat pipes?
Looks quite visible to me:

Apple didn't talk about them?

From Apple:

Engineered to go all out. All the time.
Squeezing every possible ounce of performance out of the processor means giving it a lot of power. In this case, over 300 watts. A massive heat sink keeps the system cool, enabling it to run fully unconstrained. Heat pipes move heat away from the chip, dispersing it along aluminum fin stacks. While three axial fans move air through the system.

 

[dantroline]

Doesn't really matter, since the G5's and Xeon both used the same "cheese grater" hole pattern.

But in any event, this "+20%" claim is a YAWN because the engineering details are so darn obvious: when they changed the hole pattern, they also changed the ratio of {hole diameter} -vs- {web}

(FYI "web" is the amount of metal remaining between holes).

Eyeballing it, I'd say that the web on the cMP was roughly "r" (1 radius); whereas the new case looks to be around "r/3". In simple terms, there's more hole and less metal, thus more unitized cross-sectional area to facilitate airflow...not rocket science at all.

Yes and the case is bigger, which also means it has more space in it, allowing it to be larger, etc.

 

[basehead617]

As someone who works seriously on audio, I’m concerned about noise from the display.. will be good to hear more opinions next week..