The physics and thermodynamics of the nMP and future desktops thread

Discussion in 'Mac Pro' started by Gav Mack, Dec 15, 2013.

  1. Gav Mack macrumors 68020

    Gav Mack

    Joined:
    Jun 15, 2008
    Location:
    Sagittarius A*
    #1
    I've had a quick search and not seen a thread yet.

    Despite wishing it was another big box I find the physics and thermodynamics of the black can fascinating, perhaps a game changer for a central core of the personal computer. This blown air PSU common core concept is ideal in every way and incredibly efficient so I'm looking forward to the technical reviews and tear downs from the usual places to see just how good it is in practice.

    The further part is if the efficiencies are as good as I suspect this is the beginnings of a game changer for the central element of a desktop PC. With not only 3 pcb slots on the pipe for cooling transistors but larger, 6, 9, 12 or more slots designs with standardised bus cables between the boards. IEEE please take note!

    I'll remind every can hater who can appreciate this design comment on its science now and in the future solely and put your personal feelings aside - like me :D
     
  2. AidenShaw macrumors P6

    AidenShaw

    Joined:
    Feb 8, 2003
    Location:
    The Peninsula
    #2
    I feel that many of the real advances in personal computer design have come from making standards (such as the ATX mobo) and having companies innovate designs on the standards.

    For the "quiet workstation" space, some the real quietness comes from having individually controlled fans for each heat region. The case fan spins slowly (and quietly assuming that a large diameter slow RPM fan is part of the design) until the ambient temperature in the case rises, then the case fan (only) spins up a bit. The power supply fan similarly speeds and slows according to the temperature of the PS. The CPU fan (usually a 4 wire PWM fan that can slow to barely moving) takes care of the CPU. The GPU(s) fan(s) are controlled by the GPU card.

    The Mac Mini Pro (I'm trying to avoid "trash can" references, but IMO Apple really screwed up by giving a "game changing" system exactly the same name as the previous, unrelated model) gives up a bit on this design paradigm by having a single fan that has to spin fast enough to cool the hottest region. (And since the PS doesn't seem to have a thermal connection to the core heat sink - "sharing" the cooling doesn't seem to be an option for the PS.)

    When I want to check the thermals on one of my HP servers, I log into the web server that's part of the power supply, and click "temperatures". It has 16 fans in 6 zones, and spin up a slow down as the load changes. The GUI shows the temps like:
     

    Attached Files:

  3. FrankHahn macrumors 6502a

    Joined:
    May 17, 2011
    #3
    The thermodynamics in the central cooling core of the nMP is straightforward: Air flow carries heat and hot air has a tendency to rise.

    It is a smart move in the nMP to make use of the natural phenomenon that hot air has a tendency to rise. This aids the fan in bringing the hot air up and out. The cylindrical shape eliminates the dead corners for the air flow in the rectangular shape. In the passing, air is pumped out at the back in the oMP.

    If the air flow is fast enough, then the entire system can be efficiently cooled. Abundant experiments should have been done to test whether this method of cooling is feasible in the nMP.
     
  4. ScottishCaptain macrumors 6502a

    Joined:
    Oct 4, 2008
    #4
    Who cares?

    It's all proprietary junk anyways. I've seen plenty of fan-less systems with "clever" and "innovative" cooling systems (mostly for industrial automation controllers). The nMP is no different.

    They had to cut out the disk drives, optical drives, 4 RAM slots, a CPU socket, and all the PCI-e sockets to get what they've got now. How is that impressive? 90% of that crap is just going to land up in an external enclosure now with a bunch of big old standardized DC fans cooling the lot of it. How is that revolutionary?

    The only thing "revolutionary" about the nMP is the marketing behind it, and how you're all buying it up. All Apple has succeeded in doing is throwing up their hands and saying "Not our problem anymore", because they're right- it's the consumers problem to go out and buy a swath of Thunderbolt peripherals to meet their needs, rather then buying a system that does 99% of what you want OOTB.

    So really, I'm more curious to know how much more power we're wasting on redundant AC->DC converters in external chassis, and how much MORE heat the average "pro" setup dissipates once it's been configured to be approximately as expandable and capable of the old Mac Pro.

    -SC
     
  5. Dranix macrumors 6502a

    Dranix

    Joined:
    Feb 26, 2011
    Location:
    Gelnhausen, Germany
    #5
    I don't know what you talk about. Any heavy user of MP already uses external storage systems so it's logical to have only the systemdisk in the nMP. For optical storage - they are available buspowered...
     
  6. pastrychef macrumors 601

    pastrychef

    Joined:
    Sep 15, 2006
    Location:
    New York City, NY
  7. phrehdd macrumors 68040

    phrehdd

    Joined:
    Oct 25, 2008
    #7
    I appreciate the logic of the new Mac Mini Pro but I also can say it is not anything new with respect to cooling. There have been fan units designed to handle more than one item within a computer in the past. However, I can certainly applaud the fact Apple took this path in a relatively small form.

    It will be interesting to see what happens if the fan should fail. Does the Mac Mini Pro have enough smarts to shut itself down or will it just keep going until one or more boards get fried. This is of course an all eggs in one basket scenario.
     
  8. seveej macrumors 6502a

    seveej

    Joined:
    Dec 14, 2009
    Location:
    Helsinki, Finland
    #8
    Assuming a traditional workstation design, this is quite true.
    Problematically, this "truth" is also based on the main limitation of the traditional ATX box design - the fact, that the box contains too much air, and that this air can not be transferred effectively - there are always pockets of air, which are not moved, etc. A fan is, by necessity, circular, whereas computer housing designs have (traditionally) been cubic - creating big problems. In short - you only have to look at the shape of a golfball to understand that aerodynamics is a complex interaction.

    I once saw a video on Youtube (but could not find it), which used "colored air" to visualize the airflow of a tower desktop, and it looked quite bad. The narrator of the Video made the point of noting, that the amount of dust gathered within desktop boxes (even on non-static surfaces) is testament to the relatively shabby airflow within the box.

    To make another note:AidenShaw uses the example of the GPU controlling the GPU fan, and admittedly - on the face of it - the argument makes sense. What in fact does not make sense is that the GPU has no control of it's ambient temperature and the (standard-based) limitations placed on GPU cards form factors make the optimality of GPU-controlled GPU cooling a sad joke.

    It is true, that one central cooling element (i.e. thermal core) and one fan (to rule them all) may produce problems as the cooling needs of different components vary, especially at different load profiles. Then again, most laptop users have had to contend with the setup for years, and by and large: it works.

    Thus far, we've only seen tidbits of the nMP (and I'm definitely waiting for a teardown and Anand's inquisitive eye), but for the nMP to fulfill its potential:
    - the effectiveness of thermal transfer in the thermal core (both from the contact surface to the airflow-heatsink-interface and from different parts of the thermal core) is crucial. This is largely defined by the thermal conductivity of the heatsink material and the physical form of the heatsink. I have no hard data on the material used in the heatsink, but the surprisingly low number of dissipator blades in the heatsink makes me wonder...
    - the thermal compound needs to be stellar - a topic where Apple traditionally has not been strong
    - The airflow needs to be really good, not only quantitatively but "qualitatively" as well. The ability of a heatsink to shed heat is defined as much by the temperature gradient as by the airflow profile. Traditional fans have always suffered due to the fan-engine being in the middle and thus creating an erratic airflow profile (basically creating a flow vortex "behind" the fan motor). This has always been a problem with component fans, which need to fit within a box (such as 80x80x25 mm), because simply adding a cone on top of the fan motor would significantly improve the profile of the airflow. I have yet to see anything but promo pictures, but in designing a special fan, Apple has at least had the potential to address some of the typical problems.

    We've heard PR-statements on the noise-level of the nMP, and even though the we'd take that data with a sack of salt, the promises are ... promising. Whether the results are reached through a thermal innovation or CPU/GPU throttling remains to be seen, as well as whether we'll have problems with overheating electrolytic condensers is a question for the future...

    RGDS,
     
  9. ABCDEF-Hex macrumors 6502

    ABCDEF-Hex

    Joined:
    Feb 15, 2013
    Location:
    NC
    #9
    Do you really think that they just slapped it together?

    It will be interesting to see what happens if the fan should fail. Does the Mac Mini Pro have enough smarts to shut itself down or will it just keep going until one or more boards get fried. This is of course an all eggs in one basket scenario.[/QUOTE]


    They quit building computers in a garage a long time ago.
     
  10. iBug2 macrumors 68040

    Joined:
    Jun 12, 2005
    #10
    Not at all. There's tremendous research going on mostly for aerospace engineering that tries to cut down the noise of jet engines. Friend of mine doing his Ph.D. in turbofan acoustics gave a talk 3 years ago in Paris and he said people from all major tech companies including Apple Intel and Microsoft were among the people who attended and they personally talked to him after the seminar. His specialty was jet engines on commercial aircrafts, but nevertheless, the research is applicable anywhere where something is rotating and making noise. And he said that there's a lot more that can be done on the noise issue.
     
  11. gr8tfly macrumors 603

    gr8tfly

    Joined:
    Oct 29, 2006
    Location:
    ~119W 34N
    #11
    I seem to remember either from the keynote itself one of the videos or promo material (meaning I can't remember...) that the inner core side of the heatsink had been designed to ensure a good laminar flow (probably with surface textures). This would increase the contact area between the heatsink and the coolant air, and also reduce turbulence caused by boundary area detachment. By definition, a good laminar flow will keep air moving as close as possible to the heatsink surface. Keeping turbulence to a minimum should keep air moving smoothly up through the core. Obviously, they would have looked at reducing eddy currents, and just the couple of design goals I mentioned would help.

    As some else said, they didn't build this in a garage. I'm sure Apple has access to some of the best engineers using the best tools in the industry. Hopefully, they made the correct design decisions base on their data (and the correct compromises - as there always are some).
     
  12. iBug2 macrumors 68040

    Joined:
    Jun 12, 2005
    #12
    Funny thing is, when they did build their first computer in the garage, they came up with a power supply which was smaller than anything else and more quiet. :)

    ----------



    What happens when fans fail in current towers? The same will happen with the new Mac Pro as well.

    ----------

    Since 1 out of your 2 posts about this computer references to the trash can, I don't think you are doing a good job at that :)

    Like they removed firewire, optical drive, internal hard drive, express card slot, yet they still called it Macbook Pro? Yeah, how stupid that was...
     
  13. Rich.Cohen macrumors regular

    Joined:
    Oct 28, 2013
    Location:
    Washington DC
    #13
    As I recall laminar flow is highly dependent of the cleanliness of the surface. I wonder what will happen as dust accumulates. The laminar flow will probably retard dust accumulation, but it can't stop it unless the flow is at lease occasionally at a considerable speed. I wonder if the nMP will come with instructions for regular cleaning?
     
  14. Wardenski macrumors 6502

    Joined:
    Jan 22, 2012
    #14
    The Mac Pros thermal core is an interesting concept and it does mean it can be smaller than a normal PC but aside from that I don't see any advantage. The physics will be the same for the nMP: airflow, surface area, specific heat and thermal conductivity will dictate its effectiveness.

    Some heatsinks for custom PCs are almost as large as the nMP. I was tempted by the Megahalems myself. I decided not to go ahead, I will wait for 8 core i7s.
     
  15. ScottishCaptain macrumors 6502a

    Joined:
    Oct 4, 2008
    #15
    FWIW, Thermalright (not Thermaltake!) has made heatsinks like that for years (going back to at least 2003, since that's when I started using them in my PC builds). They've always had units that supported a 120mm fan, and as such it was always possible to build a super powerful and extremely quiet machine. A friend recently used the HR22 with the optional duct and a triple-bladed Delta fan on a new PC build, and in the end his machine was quieter then my 2010 Mac Pro.

    IMHO, the "thermal core" is nothing more then an excuse for all the crap they cut out of that machine. The nMP is basically a fashion statement and very little more, which is a bloody shame. It's the return of the Cube, but this time they learned from their former mistake and discontinued the tower to make sure it'd sell.

    -SC
     
  16. Cubemmal macrumors 6502a

    Joined:
    Jun 13, 2013
    #16
    I understand your frustration but it's nothing like that. Many of us care about the features and not the cosmetics - small size, extremely quiet and low power and CPU/GPU specs. The polished tube and thermal core are being hyped per Apple marketing, but the machine itself is nothing to laugh at.

    I threw up when I first saw it at the keynote, but since then have realized that it actually was the computer I wanted. I like the oMP but frankly it was barely upgradable, I had more problems with PCIe cards (and limited selection) than not. And it's louder than I'd like - a lot louder. I want Mac Mini levels and now I'll get it, plus a Xeon brute in addition.
     
  17. ABCDEF-Hex macrumors 6502

    ABCDEF-Hex

    Joined:
    Feb 15, 2013
    Location:
    NC
    #17
    Sorry - just referring to a quote from phrehdd -

    What happens when fans fail in current towers? The same will happen with the new Mac Pro as well.

    ----------

    I was just quoting phrehdd (failed to grab his ID when I edited).
    I am going to get the nMP.
     
  18. Gav Mack thread starter macrumors 68020

    Gav Mack

    Joined:
    Jun 15, 2008
    Location:
    Sagittarius A*
    #18
    If the fan fails it won't work, I don't see Apple doing an AMD and allowing their silicon to fry like with the Athlon XP.

    This small form factor for the core of a computer with some imagination can not only be scaled to add more heatpipe bays but in future also incorporating this core into a larger enclosure to host non core components and keeping its thermodynamic design intact.

    If Apple don't expand this design after it's proved to be that efficient I can't see others not adopting similar concepts for future desktops, servers, render farms..
     
  19. Wardenski macrumors 6502

    Joined:
    Jan 22, 2012
    #19
    I don't think you can expand this concept forever. It would be funny to see a nMP with a square, pentagonal and hexagonal core etc (each one would require a custom motherboard surely) but I think there is a limit to how much heat a single heatsink can remove using just air (e.g. PowerMac G5 watercooling). I don't know what that limit is off the top of my head.

    I wonder what the effective surface area of the core is compared to the heatsinks I linked to above...
     
  20. AidenShaw macrumors P6

    AidenShaw

    Joined:
    Feb 8, 2003
    Location:
    The Peninsula
    #20
    In the cheese-grater Mac Pros, the thermal zones do exactly the same thing - eliminate the dead corners.

    Rack servers are essentially rectangular tubes, with rows of fans across the width of the tube moving the air.

    [​IMG]

    There are lots of rectangular boxes with good and quiet cooling, as well as some bad ones.

    ----------

    It's not just heat (watts), but also heat density (watts/mm²). The G5 generated a lot of heat in a fairly small chip, so the liquid cooling was used to deal with the heat density.
     
  21. OdinEng macrumors newbie

    Joined:
    Jun 22, 2013
    #21
    One fan is good

    Generally, the least reliable components in any piece of electronic equipment are the moving mechanical parts - disk drives and fans in particular. The more separate fans a piece of equipment has, the more likely it is that one of those components will fail and require shutdown and repair. So the fewer the number of fans the better, assuming one has done a careful thermal design that allows that one fan to cool all the heat loads.

    In addition, larger fans tend to move more air more efficiently than little fans. It requires no engineering to throw a bunch of little screaming fans at every heat load in a box, much more difficult to arrange the heat sinks and air flow paths within a box to allow one fan to do the job of many. But that's what Apple did.
     
  22. Gav Mack, Dec 16, 2013
    Last edited: Dec 16, 2013

    Gav Mack thread starter macrumors 68020

    Gav Mack

    Joined:
    Jun 15, 2008
    Location:
    Sagittarius A*
    #22
    I would imagine it's scalable to a dozen bays at least, more heat can be dissipated by just increasing the throughput of the coolant with a larger fan. Air pockets in heatpipes can be used to spread the heat from the hotspots to the centre of the core away from the pcb's.

    The current thermal design for desktops are a complete and utter mess efficiency wise. Take a typical PC case such as the slightly over clocked rig using a Zalman case I built for my son last month. ATX supply at bottom, CPU and GPU fans blowing in different directions and exhaust fans at front, top and rear. At full load it makes an absolute racket with air inside going all over the place. Properly designed server boxes like described previously are better but miles away from being ideal.

    A core design like this with say new IEEE standards for bay board components and interlinks to each bay would increase not only the thermal and power efficiency but cut the noise by a huge factor with a setup like I've made for my boy in the future.
     
  23. linuxcooldude macrumors 68020

    Joined:
    Mar 1, 2010
    #23
    The cooling between the nMP & rack server is still drastically different. Anything besides a smooth flat surface, will induce drag. Probably why the components were moved to the outside of the heat sink.

    The Server on the other hand, has nothing to direct the airflow and has a multitude of components on both sides that will slowdown the airflow, creating air pockets. The reason why so many air fans are needed, probably at a high rpm, which are quite noisy.
     
  24. tomvos macrumors 6502

    tomvos

    Joined:
    Jul 7, 2005
    Location:
    In the Nexus.
    #24
    I bet this won’t be a IEEE standard anytime soon. Why? Look at the Macbook Air and the Ultrabooks. The latter are blatant ripoffs from Apple’s design. I bet Apple is not amused by this ripoffs (as you can see with the Samsung lawsuits).
    So I guess, Apple would rather not see their design become a standard for everybody … just so that everybody can build cheap nMP lookalike clones with standard components.
     
  25. wildmac macrumors 65816

    Joined:
    Jun 13, 2003
    #25
    I am wondering if it's gonna get a buildup of cat hair... (I have to clean the grill on the front of my oMP about once every 2-3 weeks...)
     

Share This Page