What to expect from Apple in early 2022 - MacBook Air, Mac mini, iMac Pro, and more

1235»

Comments

  • Reply 81 of 90
    DuhSesameDuhSesame Posts: 1,278member
    Marvin said:
    tht said:
    Yes, it is the theory that Apple can have reduced retail prices for their Apple Silicon machines because they don't need to pay for Intel, AMD, and graphics memory margins with Apple Silicon. That's the theory. So, why are MBP14/16 prices higher? I think NAND is cheaper now, so there are cheaper prices from cheaper NAND GB/$ costs, but Apple is basically using the same price points for all their Apple Silicon Macs. The M1 Mac Mini is cheaper, but the M1 Pro MBP14/16 base SKUs are about $200 more than prior base SKUs. Where's this price advantage?

    An M1 MBA at $800 would sell a lot! An iMac 24 at $1000? Awesome price point. Or, why isn't the base MBP14/16 SKUs at $1800 and $2300?Even you estimate an iMac Pro with M1 Max Duo to cost $5000. Why not $4000? So, wait on see the prices. Like I said, so far, they seem to be sticking with their established price tiers give or take 10%.
    The price advantage doesn't affect entry models so much as there's a cost floor. The market for premium notebooks will also easily bear $300 chips so there's no point in selling them at $50 and making $10 profit, that's just throwing profit away. $10 x 20m units = $200m, $260 x 20m = $5.2b, there's no point in throwing away $5b when the buyer doesn't care that much. Even if unit sales went up 5x at the lower price (which they wouldn't), it's still not worth it.

    It's much the same with $800 vs $1000 laptops, people are happy to pay $200 more for premium laptops and it allows for resellers offering discounts. Some of Apple's component prices are a bit excessive. They charge $400 for 16GB RAM when choosing the Max upgrade. The old 5600M was an $800 upgrade and they only charge $400 for the M1 Max chip upgrade but force you to take the $400 RAM upgrade with it so that it's the same $800 upgrade. The good thing is the Pro model is equivalent performance to 5600M and significantly less expensive.

    I actually wrote $3,999 for the Duo model iMac Pro originally because that would be closer to the component prices but if they use more expensive memory, $4,999 is a possibility. I'm avoiding being overly optimistic. Another possibility is that they reserve the XDR display panels to higher models and charge a premium. If they have an XDR 27" iMac Pro with Max Duo for $3,999, that would be amazing.

    When comparing this to the Mac Pro style upgrades, I can't see them having a $4,999 Max Duo entry point with a selector box to M1 Max Quad priced at over $18,000, that's what they have now with Intel/AMD CPU and GPU upgrades. I reckon that selector box will at most be $3,000, possibly with a memory upgrade on top. Even with a $2k memory upgrade that's $9,999. Even with this much cheaper price, Apple still makes the same kind of profit because the difference is avoiding sending around $10,000 directly to Intel/AMD.
    tht said:
    GPUs are going into the HPC server space. All codes that solve systems of differential equations (numerical codes in physics/engineering fields) are transitioning to GPU compute. It's crazy not to as it can be a 10x performance lift. Data science and crypto are using GPU compute, at least until the specialist tensor or machine learning chips come. Apple has an advantage here with Apple Silicon as current GPUs don't have a lot of memory, which a lot of numerical codes really like. The window is closing though. One Apple Silicon 128 g-core GPU isn't enough. It's great that it will only need about 250 Watts. Well, it's even better if Apple can put 4 of them in a box.
    It's a lot of engineering effort to make that kind of box with 4 x 4-die tiles and the use case is really specialist. The people doing this kind of work can easily network 4 or more boxes together to achieve the same goal and Apple will still make 4x the sales. It helps reduce expensive refurb inventory. The other thing to consider is what else those users could be buying as an alternative. Best case would be a Threadripper plus dual 4090 around $10k. Apple's $10k option might be slower than this but it's competitive and people will think twice about getting the PC.
    tht said:
    Do you think this "Pro Workflow Team" that Ternus trots out in statements and interviews is a kind of performative art? This is in response to your "trying different strategies" comment. For some reason, they think it is comforting for potential buyers to hear it. I think it is downright frightening as it implies, if it is a forthright statement, that they didn't know how this market segment uses computers.

    I do think that if they updated the 2019 Mac Pro with Apple Silicon, they can put a M1 Max Quad in an 2-wide MPX Module and put 4 of them in a Mac Pro. I also think that the Mac mini shouldn't have a square footprint. It's just not a great shape for cooling. If it was say 12 x 6 x 1.5 inches, higher performance blower fans or axial fans could be used, and they'd probably be capable of putting anything from an M1 to M1 Max Duo in it.
    High-end computing users have different needs and there are segments who respectively loved the 2013 Mac Pro, the iMac Pro and the 2019 tower and currently still own them and are likely different portions of buyers at a similar price point with some overlap. Ideally, it would be possible to have a single form factor for all of them but I don't think that's going to be feasible.

    People's use cases are also not permanently tied to the same form factor. For a long time, musicians and video editors were accustomed to using add-in co-processor boards simply because the computers weren't fast enough. Apple even made one with the AfterBurner card. Most of these aren't needed any more, there's no need for Apple to continue selling an AfterBurner card.

    High capacity storage, fast IO connection etc are needed too so there's justification for having fast expansion capability. That can be done with a single slot and let people do what they want.
    DuhSesame said:
    I don't know what Marvin means by "SFF" but it's no trash can.  I'd say at least four double-wide slots.  On the other hand, you're not using all eight slots in the current design, at least five are occupied (I/O card, MPX modules).
    I really liked the cylinder form factor, it's a very Apple-like design, it just wasn't powerful enough, wasn't upgraded over time (supplier fault) and had no ability for users to expand on their own. If it had a single PCIe slot at the back, there wouldn't have been any problems with it. People would have dropped new GPUs and IO cards into external boxes.

    The iMac Pro probably had the potential to sell more units but it was at an even higher price point than the cylinder which cuts the potential market back down. Unit volume was likely similar to the cylinder but with more revenue. Still not enough to justify making them.

    The 2019 Mac Pro was at a higher entry point again without a display so unit volume is even lower. Apple never mentions Pro models in their earnings reports because they don't make enough revenue. If Apple made $1b per year in the Mac Pro line, that's around 100-150k units per year compared to 20-30 million Macs and there's no indication they even sell that much. HP is the biggest in the industry and they make $1.6b in workstations and it's shrinking year over year while notebooks keep growing.

    The challenge in making new workstations is figuring out what market is left and what would convince them to buy a new one. If someone owns a 2019 model like 28-core, 128GB RAM, 10TB+ internal storage, 20TFLOPs+ GPUs, what would convince them to buy a new one? It's very hard to do no matter what route they go because it's likely that model satisfies their needs and many can easily migrate down to an iMac Pro. Some video editors are migrating to M1 Max MBPs from iMacs because they are so powerful now.

    Every single Mac Pro since 2012 will be a commercial failure because there's not enough users to offset Apple's manufacturing, support and marketing investment. They are probably stuck catering to whatever enthusiast crowd remains but this used to be the GPU enthusiast crowd and there won't be GPU expansion internally. Then there's the co-processor users who don't need cards any more. That leaves people with IO needs. I think a cylinder or cube-like machine with a single slot is enough for the remaining market but I expect that the people placing the orders will be influencing the design of it. The easiest route for Apple by far is to cut the 2019 model in half and leave some PCIe slots for IO expansion.
    "Best case would be a Threadripper plus dual 4090 around $10k".

        I genuinely believe a 4-die ASi will outperform anything besides the 3990X/5990WX-level if they still have successors.  Most people (including workstation users) just think it's cool and never purchased one.  That's what the top-end is about, a showpiece.  
    Intel had 56-core Xeons but they're never released on workstations.

        As for the price, they aren't just about performance.  Your motherboard, RAM, and storage also cost a lot, for good reason.  This is where some price comparison makes no sense.  You can easily top out one or two components with a lot less, so what, everything else is inferior.  Of course, you have to pay OEMs with that hefty storage tax, though ASi will not support that much RAM and it's a whole lot faster, sounds more justifiable.


    "If it had a single PCIe slot at the back, there wouldn't have been any problems with it. People would have dropped new GPUs and IO cards into external boxes."

       The trash can is quite modular, I saw the upgrade potential but that never catches on.  With Apple Silicon the design is pointless.


    "They are probably stuck catering to whatever enthusiast crowd remains but this used to be the GPU enthusiast crowd and there won't be GPU expansion internally. Then there's the co-processor users who don't need cards any more."

         
     It seems a 4-die M1 is sufficient for this year, I wonder what a 4-die M3 will do.  It'll probably be just as competitive as four dedicated GPUs.
  • Reply 82 of 90
    DuhSesameDuhSesame Posts: 1,278member
    I think fundamentally is about which form of the workstation is the best for the next decade.  It's ironic that we always separate form from function as they're the opposite, yet it's the former that leads to innovation.  Even workstations are meant to replace minicomputers, which replaced mainframes.  Once you got enough performance, people will jump on the more available & convenient ones.

    I don't know we're going back to a trash can SFF workstation, but the scale is definitely shrinking.  Yes, we have more CPU & GPU cores, but the mainstream workstations are uniprocessors now.  HSA is ever more important, numerous accelerators will offload works that traditionally require multiple CPUs.  

    Not saying CPUs aren't important, it's still the dominant chip, you just don't need so much of it anymore.  The combination among multiple units will outperform your typical CPU-only set up by a massive amount, with significantly less power.

    Also, keep in mind that optimization for multi-core CPUs is a pain, workloads don't scale linearly with core counts, thus 64-core today is barely sold.

    With that in mind, if we can pack all that inside an SoC, we'll see a decrease in latency & power consumption, resulting in a much elegant system that can do everything the bigger ones can.  All the slots for these can go, too.  Whatever slots you have should be good for the rest of the cards, and if someone wants the main system to be upgradeable, ASi can certainly build on a module, I don't think that's hard but whether they're willing to make it happen.
  • Reply 83 of 90
    thttht Posts: 5,461member
    DuhSesame said:
    "The workstations won't be overheating because their cooling systems are designed to remove 1000 to 1500 W of heat out of the box, just like the 2019 Mac Pro is."

         Again, no they don't.  The PSU is capable of putting 1.5KW of power, though that doesn’t mean you’ll use all of that (it needs overhead, come on) or all the power will transfer to heat.  The most powerful CPU out there tops at ~300W, don't make it sound like cooling a 1000w chip is easy, if not downright impossible.  With that in mind, not every workstation needs 4GPUs, depending on what you're doing.
    Perhaps you are misunderstanding me. Yes, it means that a 1500 W workstation will have a cooling system that needs to remove the heat from 1500 W of power. I'm not talking about the CPU specifically, but the computer as a whole. The Mac Pro cooling system does this, except it's 1400 W. It supports a Xeon, 4 GPUs, and presumably 4 PCIe cards capable of using 75 W each, whatever combination thereof. All those components generate heat that needs to be transferred out of the box. "Overheating" is a derogatory term used by Marvin, at least interpreted by me as something improper when a computer is described as overheating, or perhaps he really means undesirable. Yeah, these workstations generate a lot heat. They are not overheating. They are designed to run perfectly well with the heat for years at a time. It's more than 1000 W after all. That's more than a blow dryer or a microwave or a toaster over. Like all those appliances, they run perfectly fine, and yup, they are hot.

    The only question we are arguing about is whether buyers care or not, that a small computer with little to no expansion but more power efficient will sell better than a large computer with a lot of expansion but less power efficient. Both have the same performance. My argument is that this buyers want to have as much performance as possible, and want to put 2 or 3 GPUs in it and run up the power bill. They don't really care about it being small. They just want to get their compute job as done as fast as possible.


    DuhSesame said:
    "I think people will choose a big box over a small one because the internal expansion is more valuable in this market than being small, quiet, and efficient."

         Okay, I agree.  I don't know what Marvin means by "SFF" but it's no trash can.  I'd say at least four double-wide slots.  On the other hand, you're not using all eight slots in the current design, at least five are occupied (I/O card, MPX modules).
    SFF = Small Form Factor

    The Mac mini is truly a SFF PC by most everyone's definitions, while something like a 2013 Mac Pro may not. I do view the 2013 model and the G4 Cube as SFF PCs. They are basically as small as PCs using micro ITX boards.

    Gurman says that the Apple Silicon Mac Pro will be reminiscent of the G4 cube, or bring nostalgia for the G4 cube. Prosser says the machine is like 4 or 5 Mac mini's stacked on top of each other. If so, they won't be able to support even short PCIe cards, let along have support for MPX modules.

    If it was rumored to have 4 2-wide slots, I wouldn't have much complaint.


    DuhSesame said:
    "If it was say 12 x 6 x 1.5 inches, higher performance blower fans or axial fans could be used, and they'd probably be capable of putting anything from an M1 to M1 Max Duo in it."

         Or not?  The 16" MacBook Pro had a cooling system of at least 120W.  I think it can do more, but the point is thickness aren't that restrictive.  Apple can just spin up the fan, then mini will be able to 2-die like magic.  But why do that on an entry-level system?  It won't be much cheaper, might be noisier, and potentially a lot hotter, too.  Then, why not just get an all-in-one or a portable?  Same performance, better form.
    My math is an M1 Max Duo will need about 180 W: 60 W for the CPU and 120 W for the GPU. Then you need more power for ports, other components etc.

    The Mac mini has a 150 W PSU. So, not enough for a M1 Mac Duo. Definitely enough for the M1 Max.

    The Mac mini's square footprint is not efficient for a cooling system, where the cooling fan and the radiator fins (heatsink in my vernacular) are in design opposition: making one bigger makes the other smaller. So, I don't think spinning up the fans will be enough, and it will be noisy, something Apple really doesn't want Apple Silicon machines to be.

    By making the Mac mini longer, they can can increase both the size of the fan and the heatsink. It doesn't have to be 1.5" think. It could be 1" possibly even 0.6". Being 12" long by 6" wide means something like a 5" impeller fan and a 5 x 5 x 0.5 inch heatsink for the fan to blow air across could be put in there. That should be enough for something like a 250 W machine.
    muthuk_vanalingam
  • Reply 84 of 90
    DuhSesameDuhSesame Posts: 1,278member
    tht said:
    DuhSesame said:
    "The workstations won't be overheating because their cooling systems are designed to remove 1000 to 1500 W of heat out of the box, just like the 2019 Mac Pro is."

         Again, no they don't.  The PSU is capable of putting 1.5KW of power, though that doesn’t mean you’ll use all of that (it needs overhead, come on) or all the power will transfer to heat.  The most powerful CPU out there tops at ~300W, don't make it sound like cooling a 1000w chip is easy, if not downright impossible.  With that in mind, not every workstation needs 4GPUs, depending on what you're doing.
    Perhaps you are misunderstanding me. Yes, it means that a 1500 W workstation will have a cooling system that needs to remove the heat from 1500 W of power. I'm not talking about the CPU specifically, but the computer as a whole. The Mac Pro cooling system does this, except it's 1400 W. It supports a Xeon, 4 GPUs, and presumably 4 PCIe cards capable of using 75 W each, whatever combination thereof. All those components generate heat that needs to be transferred out of the box. "Overheating" is a derogatory term used by Marvin, at least interpreted by me as something improper when a computer is described as overheating, or perhaps he really means undesirable. Yeah, these workstations generate a lot heat. They are not overheating. They are designed to run perfectly well with the heat for years at a time. It's more than 1000 W after all. That's more than a blow dryer or a microwave or a toaster over. Like all those appliances, they run perfectly fine, and yup, they are hot.

    The only question we are arguing about is whether buyers care or not, that a small computer with little to no expansion but more power efficient will sell better than a large computer with a lot of expansion but less power efficient. Both have the same performance. My argument is that this buyers want to have as much performance as possible, and want to put 2 or 3 GPUs in it and run up the power bill. They don't really care about it being small. They just want to get their compute job as done as fast as possible.


    DuhSesame said:
    "I think people will choose a big box over a small one because the internal expansion is more valuable in this market than being small, quiet, and efficient."

         Okay, I agree.  I don't know what Marvin means by "SFF" but it's no trash can.  I'd say at least four double-wide slots.  On the other hand, you're not using all eight slots in the current design, at least five are occupied (I/O card, MPX modules).
    SFF = Small Form Factor

    The Mac mini is truly a SFF PC by most everyone's definitions, while something like a 2013 Mac Pro may not. I do view the 2013 model and the G4 Cube as SFF PCs. They are basically as small as PCs using micro ITX boards.

    Gurman says that the Apple Silicon Mac Pro will be reminiscent of the G4 cube, or bring nostalgia for the G4 cube. Prosser says the machine is like 4 or 5 Mac mini's stacked on top of each other. If so, they won't be able to support even short PCIe cards, let along have support for MPX modules.

    If it was rumored to have 4 2-wide slots, I wouldn't have much complaint.


    DuhSesame said:
    "If it was say 12 x 6 x 1.5 inches, higher performance blower fans or axial fans could be used, and they'd probably be capable of putting anything from an M1 to M1 Max Duo in it."

         Or not?  The 16" MacBook Pro had a cooling system of at least 120W.  I think it can do more, but the point is thickness aren't that restrictive.  Apple can just spin up the fan, then mini will be able to 2-die like magic.  But why do that on an entry-level system?  It won't be much cheaper, might be noisier, and potentially a lot hotter, too.  Then, why not just get an all-in-one or a portable?  Same performance, better form.
    My math is an M1 Max Duo will need about 180 W: 60 W for the CPU and 120 W for the GPU. Then you need more power for ports, other components etc.

    The Mac mini has a 150 W PSU. So, not enough for a M1 Mac Duo. Definitely enough for the M1 Max.

    The Mac mini's square footprint is not efficient for a cooling system, where the cooling fan and the radiator fins (heatsink in my vernacular) are in design opposition: making one bigger makes the other smaller. So, I don't think spinning up the fans will be enough, and it will be noisy, something Apple really doesn't want Apple Silicon machines to be.

    By making the Mac mini longer, they can can increase both the size of the fan and the heatsink. It doesn't have to be 1.5" think. It could be 1" possibly even 0.6". Being 12" long by 6" wide means something like a 5" impeller fan and a 5 x 5 x 0.5 inch heatsink for the fan to blow air across could be put in there. That should be enough for something like a 250 W machine.
     "Yes, it means that a 1500 W workstation will have a cooling system that needs to remove the heat from 1500 W of power. I'm not talking about the CPU specifically, but the computer as a whole."

        First, not all of that 1.5KW needs actively cooled.  Second, using all system power just to sustain your chips is ridiculous, who'd design a system just to overload itself?  Combined CPU + 2x Duo Graphics card is about 1.1-1.2KW at best.


    "Gurman says that the Apple Silicon Mac Pro will be reminiscent of the G4 cube, or bring nostalgia for the G4 cube. Prosser says the machine is like 4 or 5 Mac mini's stacked on top of each other. If so, they won't be able to support even short PCIe cards, let along have support for MPX modules."

         I guess he also believed an 18-month upgrade cycle too, I don't trust them so much.  Use your own train of thoughts.


    "So, I don't think spinning up the fans will be enough, and it will be noisy, something Apple really doesn't want Apple Silicon machines to be."

        The new 16" is noisier under maximum load, actually, and if you concerned about PSU, put it outside.  The real reason is there's no need for a 16+4 miniature desktop, the iMac Pro/MacBook Pro will be a better choice.
    edited January 2022
  • Reply 85 of 90
    DuhSesameDuhSesame Posts: 1,278member
    Apple also have the leverage to pick & choose their dies, or actively shutting down GPU cores.  If that's not enough, since a combined scenario is extremely rare, some throttling is acceptable & you'll still get better combined performance over the competition.  It may be 180W total, but that doesn't mean you'll stick with that.
    edited January 2022
  • Reply 86 of 90
    MarvinMarvin Posts: 15,335moderator
    tht said:
    Gurman says that the Apple Silicon Mac Pro will be reminiscent of the G4 cube, or bring nostalgia for the G4 cube. Prosser says the machine is like 4 or 5 Mac mini's stacked on top of each other. If so, they won't be able to support even short PCIe cards, let along have support for MPX modules.

    If it was rumored to have 4 2-wide slots, I wouldn't have much complaint.
    There are some cube mockups:





    I don't think they'd need the handles like on the 2019 Pro as this could be carried from the bottom but it depends on how heavy the heatsink is. This wouldn't fit any full size PCIe cards but there's enough room for a connector for external PCIe expansion. Apple could make a matching 4x bay PCIe expansion box with a single connector for people who need this for things like storage, audio cards, network cards.

    Most users would be fine with this box with 2-4 storage slots allowing up to 16-32TB internal, 256GB RAM, ~500-600W with the external box allowing for optical network cards, audio cards, 100TB storage etc. Potentially AMD GPUs if they supply the drivers.
  • Reply 87 of 90
    DuhSesameDuhSesame Posts: 1,278member
    Marvin said:
    tht said:
    Gurman says that the Apple Silicon Mac Pro will be reminiscent of the G4 cube, or bring nostalgia for the G4 cube. Prosser says the machine is like 4 or 5 Mac mini's stacked on top of each other. If so, they won't be able to support even short PCIe cards, let along have support for MPX modules.

    If it was rumored to have 4 2-wide slots, I wouldn't have much complaint.
    There are some cube mockups:





    I don't think they'd need the handles like on the 2019 Pro as this could be carried from the bottom but it depends on how heavy the heatsink is. This wouldn't fit any full size PCIe cards but there's enough room for a connector for external PCIe expansion. Apple could make a matching 4x bay PCIe expansion box with a single connector for people who need this for things like storage, audio cards, network cards.

    Most users would be fine with this box with 2-4 storage slots allowing up to 16-32TB internal, 256GB RAM, ~500-600W with the external box allowing for optical network cards, audio cards, 100TB storage etc. Potentially AMD GPUs if they supply the drivers.
    Why don't we go a little bit wild?  Say, a daisy-chain solution that allows numerous expansion modules, like a huge slot underneath, turning the system into a stackable solution.  As long there's enough length, you can build an expansion bay with four double-wide slots, now it'll able meet both type of demand.


    Now come to think about it… I think the playful factor is a potential market.  Why not target that as well?  The M1/M2 is also the best embedded processor, I’d imagine a fan less M2 mini to be the perfect system in various hobbyist/industrial projects, with great performance and reliability.  Yeah, that’s a very playful computer, perfect for tinkerers and now comes with macOS.
    edited January 2022
  • Reply 88 of 90
    thttht Posts: 5,461member
    Marvin said:
    tht said:
    Yes, it is the theory that Apple can have reduced retail prices for their Apple Silicon machines because they don't need to pay for Intel, AMD, and graphics memory margins with Apple Silicon. That's the theory. So, why are MBP14/16 prices higher? I think NAND is cheaper now, so there are cheaper prices from cheaper NAND GB/$ costs, but Apple is basically using the same price points for all their Apple Silicon Macs. The M1 Mac Mini is cheaper, but the M1 Pro MBP14/16 base SKUs are about $200 more than prior base SKUs. Where's this price advantage?

    An M1 MBA at $800 would sell a lot! An iMac 24 at $1000? Awesome price point. Or, why isn't the base MBP14/16 SKUs at $1800 and $2300?Even you estimate an iMac Pro with M1 Max Duo to cost $5000. Why not $4000? So, wait on see the prices. Like I said, so far, they seem to be sticking with their established price tiers give or take 10%.
    The price advantage doesn't affect entry models so much as there's a cost floor. The market for premium notebooks will also easily bear $300 chips so there's no point in selling them at $50 and making $10 profit, that's just throwing profit away. $10 x 20m units = $200m, $260 x 20m = $5.2b, there's no point in throwing away $5b when the buyer doesn't care that much. Even if unit sales went up 5x at the lower price (which they wouldn't), it's still not worth it.

    It's much the same with $800 vs $1000 laptops, people are happy to pay $200 more for premium laptops and it allows for resellers offering discounts. Some of Apple's component prices are a bit excessive. They charge $400 for 16GB RAM when choosing the Max upgrade. The old 5600M was an $800 upgrade and they only charge $400 for the M1 Max chip upgrade but force you to take the $400 RAM upgrade with it so that it's the same $800 upgrade. The good thing is the Pro model is equivalent performance to 5600M and significantly less expensive.

    I actually wrote $3,999 for the Duo model iMac Pro originally because that would be closer to the component prices but if they use more expensive memory, $4,999 is a possibility. I'm avoiding being overly optimistic. Another possibility is that they reserve the XDR display panels to higher models and charge a premium. If they have an XDR 27" iMac Pro with Max Duo for $3,999, that would be amazing.

    When comparing this to the Mac Pro style upgrades, I can't see them having a $4,999 Max Duo entry point with a selector box to M1 Max Quad priced at over $18,000, that's what they have now with Intel/AMD CPU and GPU upgrades. I reckon that selector box will at most be $3,000, possibly with a memory upgrade on top. Even with a $2k memory upgrade that's $9,999. Even with this much cheaper price, Apple still makes the same kind of profit because the difference is avoiding sending around $10,000 directly to Intel/AMD.
    Regarding entry models and cost floors, I think I can counter one turtle up regarding the MBP14/16 prices. Ie, a MBP14 with an 8+2+16 M1 Pro (a fully featured chip) costs $2300. Why not $2000? I think you should also question whether Apple Silicon is really cheaper for Apple than buying chips from merchant vendors. It could be that an M1 Pro chip and M1 Max chip (including its LPDRR5 memory) is just as expensive as or more expensive than an Intel CPU, its PCH, the GPU and its memory. The M1 Max is a >400 mm2 chip. That's quite big as client chips go.

    Either way, Apple's price structures are developed by starting with a price for a device and then putting in the features that would justify the price, give or take 10% pending on economies of scale. And for the most part, they do not want to have iMacs to have price overlap with other headless desktop machines at least with the SKUs. The presumed iMac 27 is going to range from $2000 to $4000. Per not wanting overlap, I think an iMac with a M1 Max Duo for $5000 is an easy call. A SFF Mac Pro? Who knows at this point. If it was like a 2019 Mac Pro, $6000 is easy call as people would pay for the slots and the expansion, assuming they can be expanded with Apple Silicon MPX modules.

    An MBP16 with an M1 Max (8+2+32), 32 GB of RAM and 512 GB of NAND storage is $3300. A similar config in terms of performance (to a M1 Max) is an iMac 5K with Intel 10c i9 and Radeon Pro 5700XT. Basically the top end iMac 5K in terms of performance. This is $3800. I think adding another M1 Max for a dual config and an 27" miniLED makes it $5000 easy, especially if they have 64 GB of memory with it.

    Currently, I think they will want to have a Mac Pro with M1 Max be priced above that iMac SKU, so $6000. Whatever needs to be in the box for people to "ok" with it at that price, they'll put in. 8 PCIe slots and the flexibility it afforded was enough. The rumors are saying it will be a small box. So who knows. Maybe 64 GB RAM and 1 TB NAND will be the minimum. Then having 2 M1 Max chips in a package presents a rather huge upgrade. It's 2x the CPU cores, 2x GPU cores, 2x the memory channels, and 2x the memory capacity. They are going to charge as much as they possibly can for that. Look at the Mac Pro base SKU. Double everything: double the CPU cores, double the GPUs, double the RAM. You get to over 10K in a hurry.


    Marvin said:
    tht said:
    GPUs are going into the HPC server space. All codes that solve systems of differential equations (numerical codes in physics/engineering fields) are transitioning to GPU compute. It's crazy not to as it can be a 10x performance lift. Data science and crypto are using GPU compute, at least until the specialist tensor or machine learning chips come. Apple has an advantage here with Apple Silicon as current GPUs don't have a lot of memory, which a lot of numerical codes really like. The window is closing though. One Apple Silicon 128 g-core GPU isn't enough. It's great that it will only need about 250 Watts. Well, it's even better if Apple can put 4 of them in a box.
    It's a lot of engineering effort to make that kind of box with 4 x 4-die tiles and the use case is really specialist. The people doing this kind of work can easily network 4 or more boxes together to achieve the same goal and Apple will still make 4x the sales. It helps reduce expensive refurb inventory. The other thing to consider is what else those users could be buying as an alternative. Best case would be a Threadripper plus dual 4090 around $10k. Apple's $10k option might be slower than this but it's competitive and people will think twice about getting the PC.
    I'm not saying putting 16 chips in a package (4 x 4 tiles). I'm saying put an M1 Max Quad in an MPX module. That's just using the same M1 Max Quad, but you can put 1 more, 2 more, 4 more in the box depending on how many MPX modules can be put in there.


    Marvin said:
    tht said:
    Do you think this "Pro Workflow Team" that Ternus trots out in statements and interviews is a kind of performative art? This is in response to your "trying different strategies" comment. For some reason, they think it is comforting for potential buyers to hear it. I think it is downright frightening as it implies, if it is a forthright statement, that they didn't know how this market segment uses computers.

    I do think that if they updated the 2019 Mac Pro with Apple Silicon, they can put a M1 Max Quad in an 2-wide MPX Module and put 4 of them in a Mac Pro. I also think that the Mac mini shouldn't have a square footprint. It's just not a great shape for cooling. If it was say 12 x 6 x 1.5 inches, higher performance blower fans or axial fans could be used, and they'd probably be capable of putting anything from an M1 to M1 Max Duo in it.
    High-end computing users have different needs and there are segments who respectively loved the 2013 Mac Pro, the iMac Pro and the 2019 tower and currently still own them and are likely different portions of buyers at a similar price point with some overlap. Ideally, it would be possible to have a single form factor for all of them but I don't think that's going to be feasible.

    People's use cases are also not permanently tied to the same form factor. For a long time, musicians and video editors were accustomed to using add-in co-processor boards simply because the computers weren't fast enough. Apple even made one with the AfterBurner card. Most of these aren't needed any more, there's no need for Apple to continue selling an AfterBurner card.

    High capacity storage, fast IO connection etc are needed too so there's justification for having fast expansion capability. That can be done with a single slot and let people do what they want.
    Yup. People, not just high end computer users, have different needs. Apple doesn't have to have a simple lineup of PCs. They can have a lineup of headless desktops spanning $500 to $5000, a lineup of AIOs spanning $1000 to $4000. a lineup of laptops spanning $800 to $3000. It doesn't have to be a ridiculous sprawl of models, but something more varied than today. It's really only about 2 or 3 more models than today. They need a headless desktop in-between a Mac mini and a Mac Pro for about $2000, including room for about 2 3.5" HDDs and about 3 PCIe slots. I think they need an M1 MBA15 at about $1800 to $2000 starting price. I think they really should have an 13" laptop at $800 with education deals at $600 yo $700. Iffy on a workstation laptop with an M1 Max Duo. They really don't have to have some people grudgingly accept what they are offering, while having a few more models would make most everyone much happier, and be able to sell more units. (Maybe Services will convince them that it is viable as they could be getting that continuous subscription money with a lot of buyers?)

    A lot of people hate that they need have multiple external storage boxes, PCIe breakout boxes, port dongles, whatever else box. It makes for a mess of cabling, and power extension cables with just the right outlet spacing because everyone must orient their power bricks differently. It presents as an uncomfortable mess on a desk.

    Your comments on what Apple can do with a SFF Mac Pro: breakout boxes et al are the same comments that we had for the 2013 Mac Pro. Firstly, it doesn't have to be like Highlander. Apple doesn't have to have one type of machine. It's actually rumored that they aren't going to have 1 machine as the 2019 Mac Pro will get an Ice Lake update, and maybe that means it will have an Apple Silicon update in 2024, but uh, that's a long time to wait, and there has been not much rumor on the 2019 Mac Pro having an Apple Silicon update at all. Anyways I digress, if they make an Apple Silicon SFF Mac Pro and an Apple Silicon big box Mac Pro, nobody argues. They just get what they want.

    Secondly, if they are going to rely on external expansion for a SFF Mac Pro, they need to it themselves and hopefully have it nicely integrated. They need to build a PCIe breakout box. They need to build a RAID box. They need to build a monitor. And mostly importantly, update it, preferably yearly.
    muthuk_vanalingam
  • Reply 89 of 90
    DuhSesameDuhSesame Posts: 1,278member
    tht said:
    Marvin said:
    tht said:
    Yes, it is the theory that Apple can have reduced retail prices for their Apple Silicon machines because they don't need to pay for Intel, AMD, and graphics memory margins with Apple Silicon. That's the theory. So, why are MBP14/16 prices higher? I think NAND is cheaper now, so there are cheaper prices from cheaper NAND GB/$ costs, but Apple is basically using the same price points for all their Apple Silicon Macs. The M1 Mac Mini is cheaper, but the M1 Pro MBP14/16 base SKUs are about $200 more than prior base SKUs. Where's this price advantage?

    An M1 MBA at $800 would sell a lot! An iMac 24 at $1000? Awesome price point. Or, why isn't the base MBP14/16 SKUs at $1800 and $2300?Even you estimate an iMac Pro with M1 Max Duo to cost $5000. Why not $4000? So, wait on see the prices. Like I said, so far, they seem to be sticking with their established price tiers give or take 10%.
    The price advantage doesn't affect entry models so much as there's a cost floor. The market for premium notebooks will also easily bear $300 chips so there's no point in selling them at $50 and making $10 profit, that's just throwing profit away. $10 x 20m units = $200m, $260 x 20m = $5.2b, there's no point in throwing away $5b when the buyer doesn't care that much. Even if unit sales went up 5x at the lower price (which they wouldn't), it's still not worth it.

    It's much the same with $800 vs $1000 laptops, people are happy to pay $200 more for premium laptops and it allows for resellers offering discounts. Some of Apple's component prices are a bit excessive. They charge $400 for 16GB RAM when choosing the Max upgrade. The old 5600M was an $800 upgrade and they only charge $400 for the M1 Max chip upgrade but force you to take the $400 RAM upgrade with it so that it's the same $800 upgrade. The good thing is the Pro model is equivalent performance to 5600M and significantly less expensive.

    I actually wrote $3,999 for the Duo model iMac Pro originally because that would be closer to the component prices but if they use more expensive memory, $4,999 is a possibility. I'm avoiding being overly optimistic. Another possibility is that they reserve the XDR display panels to higher models and charge a premium. If they have an XDR 27" iMac Pro with Max Duo for $3,999, that would be amazing.

    When comparing this to the Mac Pro style upgrades, I can't see them having a $4,999 Max Duo entry point with a selector box to M1 Max Quad priced at over $18,000, that's what they have now with Intel/AMD CPU and GPU upgrades. I reckon that selector box will at most be $3,000, possibly with a memory upgrade on top. Even with a $2k memory upgrade that's $9,999. Even with this much cheaper price, Apple still makes the same kind of profit because the difference is avoiding sending around $10,000 directly to Intel/AMD.
    Regarding entry models and cost floors, I think I can counter one turtle up regarding the MBP14/16 prices. Ie, a MBP14 with an 8+2+16 M1 Pro (a fully featured chip) costs $2300. Why not $2000? I think you should also question whether Apple Silicon is really cheaper for Apple than buying chips from merchant vendors. It could be that an M1 Pro chip and M1 Max chip (including its LPDRR5 memory) is just as expensive as or more expensive than an Intel CPU, its PCH, the GPU and its memory. The M1 Max is a >400 mm2 chip. That's quite big as client chips go.

    Either way, Apple's price structures are developed by starting with a price for a device and then putting in the features that would justify the price, give or take 10% pending on economies of scale. And for the most part, they do not want to have iMacs to have price overlap with other headless desktop machines at least with the SKUs. The presumed iMac 27 is going to range from $2000 to $4000. Per not wanting overlap, I think an iMac with a M1 Max Duo for $5000 is an easy call. A SFF Mac Pro? Who knows at this point. If it was like a 2019 Mac Pro, $6000 is easy call as people would pay for the slots and the expansion, assuming they can be expanded with Apple Silicon MPX modules.

    An MBP16 with an M1 Max (8+2+32), 32 GB of RAM and 512 GB of NAND storage is $3300. A similar config in terms of performance (to a M1 Max) is an iMac 5K with Intel 10c i9 and Radeon Pro 5700XT. Basically the top end iMac 5K in terms of performance. This is $3800. I think adding another M1 Max for a dual config and an 27" miniLED makes it $5000 easy, especially if they have 64 GB of memory with it.

    Currently, I think they will want to have a Mac Pro with M1 Max be priced above that iMac SKU, so $6000. Whatever needs to be in the box for people to "ok" with it at that price, they'll put in. 8 PCIe slots and the flexibility it afforded was enough. The rumors are saying it will be a small box. So who knows. Maybe 64 GB RAM and 1 TB NAND will be the minimum. Then having 2 M1 Max chips in a package presents a rather huge upgrade. It's 2x the CPU cores, 2x GPU cores, 2x the memory channels, and 2x the memory capacity. They are going to charge as much as they possibly can for that. Look at the Mac Pro base SKU. Double everything: double the CPU cores, double the GPUs, double the RAM. You get to over 10K in a hurry.


    Marvin said:
    tht said:
    GPUs are going into the HPC server space. All codes that solve systems of differential equations (numerical codes in physics/engineering fields) are transitioning to GPU compute. It's crazy not to as it can be a 10x performance lift. Data science and crypto are using GPU compute, at least until the specialist tensor or machine learning chips come. Apple has an advantage here with Apple Silicon as current GPUs don't have a lot of memory, which a lot of numerical codes really like. The window is closing though. One Apple Silicon 128 g-core GPU isn't enough. It's great that it will only need about 250 Watts. Well, it's even better if Apple can put 4 of them in a box.
    It's a lot of engineering effort to make that kind of box with 4 x 4-die tiles and the use case is really specialist. The people doing this kind of work can easily network 4 or more boxes together to achieve the same goal and Apple will still make 4x the sales. It helps reduce expensive refurb inventory. The other thing to consider is what else those users could be buying as an alternative. Best case would be a Threadripper plus dual 4090 around $10k. Apple's $10k option might be slower than this but it's competitive and people will think twice about getting the PC.
    I'm not saying putting 16 chips in a package (4 x 4 tiles). I'm saying put an M1 Max Quad in an MPX module. That's just using the same M1 Max Quad, but you can put 1 more, 2 more, 4 more in the box depending on how many MPX modules can be put in there.


    Marvin said:
    tht said:
    Do you think this "Pro Workflow Team" that Ternus trots out in statements and interviews is a kind of performative art? This is in response to your "trying different strategies" comment. For some reason, they think it is comforting for potential buyers to hear it. I think it is downright frightening as it implies, if it is a forthright statement, that they didn't know how this market segment uses computers.

    I do think that if they updated the 2019 Mac Pro with Apple Silicon, they can put a M1 Max Quad in an 2-wide MPX Module and put 4 of them in a Mac Pro. I also think that the Mac mini shouldn't have a square footprint. It's just not a great shape for cooling. If it was say 12 x 6 x 1.5 inches, higher performance blower fans or axial fans could be used, and they'd probably be capable of putting anything from an M1 to M1 Max Duo in it.
    High-end computing users have different needs and there are segments who respectively loved the 2013 Mac Pro, the iMac Pro and the 2019 tower and currently still own them and are likely different portions of buyers at a similar price point with some overlap. Ideally, it would be possible to have a single form factor for all of them but I don't think that's going to be feasible.

    People's use cases are also not permanently tied to the same form factor. For a long time, musicians and video editors were accustomed to using add-in co-processor boards simply because the computers weren't fast enough. Apple even made one with the AfterBurner card. Most of these aren't needed any more, there's no need for Apple to continue selling an AfterBurner card.

    High capacity storage, fast IO connection etc are needed too so there's justification for having fast expansion capability. That can be done with a single slot and let people do what they want.
    Yup. People, not just high end computer users, have different needs. Apple doesn't have to have a simple lineup of PCs. They can have a lineup of headless desktops spanning $500 to $5000, a lineup of AIOs spanning $1000 to $4000. a lineup of laptops spanning $800 to $3000. It doesn't have to be a ridiculous sprawl of models, but something more varied than today. It's really only about 2 or 3 more models than today. They need a headless desktop in-between a Mac mini and a Mac Pro for about $2000, including room for about 2 3.5" HDDs and about 3 PCIe slots. I think they need an M1 MBA15 at about $1800 to $2000 starting price. I think they really should have an 13" laptop at $800 with education deals at $600 yo $700. Iffy on a workstation laptop with an M1 Max Duo. They really don't have to have some people grudgingly accept what they are offering, while having a few more models would make most everyone much happier, and be able to sell more units. (Maybe Services will convince them that it is viable as they could be getting that continuous subscription money with a lot of buyers?)

    A lot of people hate that they need have multiple external storage boxes, PCIe breakout boxes, port dongles, whatever else box. It makes for a mess of cabling, and power extension cables with just the right outlet spacing because everyone must orient their power bricks differently. It presents as an uncomfortable mess on a desk.

    Your comments on what Apple can do with a SFF Mac Pro: breakout boxes et al are the same comments that we had for the 2013 Mac Pro. Firstly, it doesn't have to be like Highlander. Apple doesn't have to have one type of machine. It's actually rumored that they aren't going to have 1 machine as the 2019 Mac Pro will get an Ice Lake update, and maybe that means it will have an Apple Silicon update in 2024, but uh, that's a long time to wait, and there has been not much rumor on the 2019 Mac Pro having an Apple Silicon update at all. Anyways I digress, if they make an Apple Silicon SFF Mac Pro and an Apple Silicon big box Mac Pro, nobody argues. They just get what they want.

    Secondly, if they are going to rely on external expansion for a SFF Mac Pro, they need to it themselves and hopefully have it nicely integrated. They need to build a PCIe breakout box. They need to build a RAID box. They need to build a monitor. And mostly importantly, update it, preferably yearly.
    Why don’t we continue on my concepts?  I think as long there’s a stable connection, stackable modules might be very interesting.  The system itself need to be enough to accommodate various PCIe cards, then the slot underneath is capable for 64 PCIe lanes, regardless the revision.

    That, along with a passively-cooled M2 Mac mini, the perfect embedded Mac system, this new Apple Silicon lineup is very, very playful.
  • Reply 90 of 90
    DuhSesameDuhSesame Posts: 1,278member
    tht said:
    Marvin said:
    tht said:
    Yes, it is the theory that Apple can have reduced retail prices for their Apple Silicon machines because they don't need to pay for Intel, AMD, and graphics memory margins with Apple Silicon. That's the theory. So, why are MBP14/16 prices higher? I think NAND is cheaper now, so there are cheaper prices from cheaper NAND GB/$ costs, but Apple is basically using the same price points for all their Apple Silicon Macs. The M1 Mac Mini is cheaper, but the M1 Pro MBP14/16 base SKUs are about $200 more than prior base SKUs. Where's this price advantage?

    An M1 MBA at $800 would sell a lot! An iMac 24 at $1000? Awesome price point. Or, why isn't the base MBP14/16 SKUs at $1800 and $2300?Even you estimate an iMac Pro with M1 Max Duo to cost $5000. Why not $4000? So, wait on see the prices. Like I said, so far, they seem to be sticking with their established price tiers give or take 10%.
    The price advantage doesn't affect entry models so much as there's a cost floor. The market for premium notebooks will also easily bear $300 chips so there's no point in selling them at $50 and making $10 profit, that's just throwing profit away. $10 x 20m units = $200m, $260 x 20m = $5.2b, there's no point in throwing away $5b when the buyer doesn't care that much. Even if unit sales went up 5x at the lower price (which they wouldn't), it's still not worth it.

    It's much the same with $800 vs $1000 laptops, people are happy to pay $200 more for premium laptops and it allows for resellers offering discounts. Some of Apple's component prices are a bit excessive. They charge $400 for 16GB RAM when choosing the Max upgrade. The old 5600M was an $800 upgrade and they only charge $400 for the M1 Max chip upgrade but force you to take the $400 RAM upgrade with it so that it's the same $800 upgrade. The good thing is the Pro model is equivalent performance to 5600M and significantly less expensive.

    I actually wrote $3,999 for the Duo model iMac Pro originally because that would be closer to the component prices but if they use more expensive memory, $4,999 is a possibility. I'm avoiding being overly optimistic. Another possibility is that they reserve the XDR display panels to higher models and charge a premium. If they have an XDR 27" iMac Pro with Max Duo for $3,999, that would be amazing.

    When comparing this to the Mac Pro style upgrades, I can't see them having a $4,999 Max Duo entry point with a selector box to M1 Max Quad priced at over $18,000, that's what they have now with Intel/AMD CPU and GPU upgrades. I reckon that selector box will at most be $3,000, possibly with a memory upgrade on top. Even with a $2k memory upgrade that's $9,999. Even with this much cheaper price, Apple still makes the same kind of profit because the difference is avoiding sending around $10,000 directly to Intel/AMD.
    Regarding entry models and cost floors, I think I can counter one turtle up regarding the MBP14/16 prices. Ie, a MBP14 with an 8+2+16 M1 Pro (a fully featured chip) costs $2300. Why not $2000? I think you should also question whether Apple Silicon is really cheaper for Apple than buying chips from merchant vendors. It could be that an M1 Pro chip and M1 Max chip (including its LPDRR5 memory) is just as expensive as or more expensive than an Intel CPU, its PCH, the GPU and its memory. The M1 Max is a >400 mm2 chip. That's quite big as client chips go.

    Either way, Apple's price structures are developed by starting with a price for a device and then putting in the features that would justify the price, give or take 10% pending on economies of scale. And for the most part, they do not want to have iMacs to have price overlap with other headless desktop machines at least with the SKUs. The presumed iMac 27 is going to range from $2000 to $4000. Per not wanting overlap, I think an iMac with a M1 Max Duo for $5000 is an easy call. A SFF Mac Pro? Who knows at this point. If it was like a 2019 Mac Pro, $6000 is easy call as people would pay for the slots and the expansion, assuming they can be expanded with Apple Silicon MPX modules.

    An MBP16 with an M1 Max (8+2+32), 32 GB of RAM and 512 GB of NAND storage is $3300. A similar config in terms of performance (to a M1 Max) is an iMac 5K with Intel 10c i9 and Radeon Pro 5700XT. Basically the top end iMac 5K in terms of performance. This is $3800. I think adding another M1 Max for a dual config and an 27" miniLED makes it $5000 easy, especially if they have 64 GB of memory with it.

    Currently, I think they will want to have a Mac Pro with M1 Max be priced above that iMac SKU, so $6000. Whatever needs to be in the box for people to "ok" with it at that price, they'll put in. 8 PCIe slots and the flexibility it afforded was enough. The rumors are saying it will be a small box. So who knows. Maybe 64 GB RAM and 1 TB NAND will be the minimum. Then having 2 M1 Max chips in a package presents a rather huge upgrade. It's 2x the CPU cores, 2x GPU cores, 2x the memory channels, and 2x the memory capacity. They are going to charge as much as they possibly can for that. Look at the Mac Pro base SKU. Double everything: double the CPU cores, double the GPUs, double the RAM. You get to over 10K in a hurry.


    Marvin said:
    tht said:
    GPUs are going into the HPC server space. All codes that solve systems of differential equations (numerical codes in physics/engineering fields) are transitioning to GPU compute. It's crazy not to as it can be a 10x performance lift. Data science and crypto are using GPU compute, at least until the specialist tensor or machine learning chips come. Apple has an advantage here with Apple Silicon as current GPUs don't have a lot of memory, which a lot of numerical codes really like. The window is closing though. One Apple Silicon 128 g-core GPU isn't enough. It's great that it will only need about 250 Watts. Well, it's even better if Apple can put 4 of them in a box.
    It's a lot of engineering effort to make that kind of box with 4 x 4-die tiles and the use case is really specialist. The people doing this kind of work can easily network 4 or more boxes together to achieve the same goal and Apple will still make 4x the sales. It helps reduce expensive refurb inventory. The other thing to consider is what else those users could be buying as an alternative. Best case would be a Threadripper plus dual 4090 around $10k. Apple's $10k option might be slower than this but it's competitive and people will think twice about getting the PC.
    I'm not saying putting 16 chips in a package (4 x 4 tiles). I'm saying put an M1 Max Quad in an MPX module. That's just using the same M1 Max Quad, but you can put 1 more, 2 more, 4 more in the box depending on how many MPX modules can be put in there.


    Marvin said:
    tht said:
    Do you think this "Pro Workflow Team" that Ternus trots out in statements and interviews is a kind of performative art? This is in response to your "trying different strategies" comment. For some reason, they think it is comforting for potential buyers to hear it. I think it is downright frightening as it implies, if it is a forthright statement, that they didn't know how this market segment uses computers.

    I do think that if they updated the 2019 Mac Pro with Apple Silicon, they can put a M1 Max Quad in an 2-wide MPX Module and put 4 of them in a Mac Pro. I also think that the Mac mini shouldn't have a square footprint. It's just not a great shape for cooling. If it was say 12 x 6 x 1.5 inches, higher performance blower fans or axial fans could be used, and they'd probably be capable of putting anything from an M1 to M1 Max Duo in it.
    High-end computing users have different needs and there are segments who respectively loved the 2013 Mac Pro, the iMac Pro and the 2019 tower and currently still own them and are likely different portions of buyers at a similar price point with some overlap. Ideally, it would be possible to have a single form factor for all of them but I don't think that's going to be feasible.

    People's use cases are also not permanently tied to the same form factor. For a long time, musicians and video editors were accustomed to using add-in co-processor boards simply because the computers weren't fast enough. Apple even made one with the AfterBurner card. Most of these aren't needed any more, there's no need for Apple to continue selling an AfterBurner card.

    High capacity storage, fast IO connection etc are needed too so there's justification for having fast expansion capability. That can be done with a single slot and let people do what they want.
    Yup. People, not just high end computer users, have different needs. Apple doesn't have to have a simple lineup of PCs. They can have a lineup of headless desktops spanning $500 to $5000, a lineup of AIOs spanning $1000 to $4000. a lineup of laptops spanning $800 to $3000. It doesn't have to be a ridiculous sprawl of models, but something more varied than today. It's really only about 2 or 3 more models than today. They need a headless desktop in-between a Mac mini and a Mac Pro for about $2000, including room for about 2 3.5" HDDs and about 3 PCIe slots. I think they need an M1 MBA15 at about $1800 to $2000 starting price. I think they really should have an 13" laptop at $800 with education deals at $600 yo $700. Iffy on a workstation laptop with an M1 Max Duo. They really don't have to have some people grudgingly accept what they are offering, while having a few more models would make most everyone much happier, and be able to sell more units. (Maybe Services will convince them that it is viable as they could be getting that continuous subscription money with a lot of buyers?)

    A lot of people hate that they need have multiple external storage boxes, PCIe breakout boxes, port dongles, whatever else box. It makes for a mess of cabling, and power extension cables with just the right outlet spacing because everyone must orient their power bricks differently. It presents as an uncomfortable mess on a desk.

    Your comments on what Apple can do with a SFF Mac Pro: breakout boxes et al are the same comments that we had for the 2013 Mac Pro. Firstly, it doesn't have to be like Highlander. Apple doesn't have to have one type of machine. It's actually rumored that they aren't going to have 1 machine as the 2019 Mac Pro will get an Ice Lake update, and maybe that means it will have an Apple Silicon update in 2024, but uh, that's a long time to wait, and there has been not much rumor on the 2019 Mac Pro having an Apple Silicon update at all. Anyways I digress, if they make an Apple Silicon SFF Mac Pro and an Apple Silicon big box Mac Pro, nobody argues. They just get what they want.

    Secondly, if they are going to rely on external expansion for a SFF Mac Pro, they need to it themselves and hopefully have it nicely integrated. They need to build a PCIe breakout box. They need to build a RAID box. They need to build a monitor. And mostly importantly, update it, preferably yearly.
    Just have an argument.  Design aside, I think the very first priority should be bringing down the cost, at least the entry to mid-level.  The current model is not the machine that serves you and me, but for them, and the cost is no factor in that market.
Sign In or Register to comment.