The Neural unit very probably won't be optional on any system that Apple is going to want to add Face ID to. ( laptops plus probably all iMacs. ) [ Wouldn't be surprising if Apple added TouchID power button to the "headless models" too so... pretty Nueral for all is a good chance. Same way T2 in systems that don't have touchbar. Just easier to incur the cost over the whole Mac product line. ]
GPUs probably at present in all. The requirement to urn unmodified iPhone apps probably going to push at least a "good" (not necessarily best) iPhone GPU into the higher end desktop units also. Not necessarily trying to displace a more hefty discrete GPU for Mac targeted software, but something to fall back the 'clueless not running in iOS' apps to. ( variation on the trick the touchbar uses and just copy the framebuffer of of the apple GPU onto whatever window frame the users is using if engaged a dGPU to drive monitor).
An Apple GPU every SoC package means that even if pull all the discrete GPUs from a Mac Pro could still drive the TBv2 ports. ( similar to the current 256GB entry SSD, just has to be big enough to get the basics done on a entry , 'stripped down' model. )
The Mini's problem is that it doesn't have enough GPU . It doesn't need "smaller" it needs bigger to better hold more ground.
Apple's Neural Engine is likely going to be anchored on Apple "AI" tasks far more so than broad general market , leading edge performance metrics. They'll let other apps using it so their apps can do smart inference but it is probably not going to be a marketed at hard core AI/ML computational needs.
The bigger gap is going to be in I/O bandwidth ( and bisection internal bandwidth). Both memory and PCI-e I/O. There is a decent possibly they will abandoned the 1.5TB RAM capacity level. Also doubtful they will try to drive anywhere near the 64 core cap that macOS lays on them that they probably aren't going to cross for the foressable future. Would be surprising if they did anything over 32 cores. Matching the 28 core count limit probably will be "good enough" on performance cores. Apple will get a 'win' because the 2022 tech being used if better than the pragmatically 2017 tech being used in the Mac Pro. They are probably not going to get entangled in the core count wars of 2022 at the high end workstation level.
[automerge]1596132009[/automerge]
13" Retina MBP AS $1,299
14" Micro-LED MBP AS $1,899
That $600 gap means can have two products that don't pragmatically overlap. Apple sold non Retina along side Retina MBP versions at some point of the same screen dimensions. They also sold matte vs gloss MBP 15" versions along side each other in the past too.
Different tech in screen at the same size still can be differentiated.
The other question is whether Apple is going to consolidate their two "two port" options. MBP 13" two port and MBA into one . Maybe bring back some other lower cost option as "Macbook". Again the 14" could be the four port model.
Depending upon where Apple SoC is at with being able to deal with discrete Thunderbolt Controllers a MBP 13" with just one TBv4 controller (2 ports ) and a current Retina screen would be a substantively lower 'new, volatile tech' hurdle for Apple to leap over for a first product.
Pretty good chance that Apple is in way their head in trying to revamp the whole line up at the same time. Add a limited number of new things at the same time and iterate or the whole line up over 2 years. Otherwise they likely will likely have a "butterfly keyboard" like hole in the ground.
Neither industrial design or The silicon team have shown real capacity to work on large number of products in parallel. ( in part Apple didn't do all of the "Big Bang" of 2006 on their own either without shoveling board work out to Intel for the Mac Pro. When pull it all inside there isn't going to be a fall back external vendor to kick it too. )
Last edited:
