Whispers...
One thing that is necessary to be done is to release the 7,1 asap.
The second one is to have a new one after 2-3 years max.
....
It will be very disappointing, at least, to have a need for a new one after 4 years and it is still the same 2019 model, again.
If Apple fails on that second point. It probably won't be 'disappointing' after 4 years ( 2023). More likely the product would be comatose at best or a zombie at worst. If Apple is counting on some "just large enough" , rabid Mac Pro cult to hang around for a 3rd round of Rip van Winkle exercise, then I suspect they are the ones that are going to be disappointed.
Herding an increasingly shrinking pool of users into a longer and longer upgrade cycles is most likely a death spiral. They'll loose users on each iteration. That leads to longer cycles to herd into critical mass. Which leads to smaller pool. rise and repeat. For Apple is a cash cow the first couple iterations but at some point it tends to turn into musical chairs.
As for "asap" (ass soon as possible ), the critical factor is the pragmatic 'possible' given whatever plan and resource assignments Apple came up with 1-2 years ago. If Apple 'bet the farm' on some tech that is sliding out to Q4 .... it is very possible they'll skip saying anything substantive at WWDC also. ( they aren't very likely to come up with a "dog ate our homework' story to be weaved into WWDC.... whatever Cirque de Soleil gyration spin they want to put on it. )
[doublepost=1556817465][/doublepost]
It could be that Apple had to rethink T3 chip in order to get it work on a (more) modular computer.
The Mac Mni is modular ( in same sense that it needs a Display via the context Apple has mentioned Mac Pro modular) and it has a T2. It works.
Apple needs a T-series that works
across the whole Mac line up to get reasonable scale for the T-series ( make it lower cost. ).
Time on what? ......
If there are PCIe dGPU slots in mMP, T3 could verify their Apple approval in a boot process.
That isn't what the T-series does. The T2 validates software/firmware ; not hardware. There is about zero usefulness in modifying the T-series silicon to exclude some piece of hardware. That could be all done in firmware. But that is extremely uncharacteristic of Apple's approach to software in the past. if Apple has a specific boot GPU they want that'd would be one thing. But if they do that ( e.g., Thunderbolt integration) it is not particularly likely that it would be a generic PCI-e standard slot spec. There would be a PCI-e component to the socket/slot but Apple could simply add some physical mods so a random card just can't be placed there. Done. No super duper silicon circuit modes needed.
Apple is shutting down openGL and openCL and denying 32bit apps.
Which has zero overlap with T2's primary functions ( firmware validation , SSD controller, SMC controller , Audio/Video mic/'facetime cam' handling).
Could be that Apple is going to do a bigger kernel level change soon,
If there is a new driver API then some kernel changes are probably happening. That might be tangentially related to making the macos/T2 software bridge interface more stable. But that wouldn't necessarily be a silicon thing.
iOS and macOS already share a kernel. So there probably isn't going to be an "iOSisfication" of the kernel. Apple may roll out some cleanup merges to 'unfork" some drift that may have built up over time, but that's probably going to be hyped out of proportion ( although it probably will break some things so 'dead' and comatose drivers will break. ).
That really doesn't make much sense. The primary point of having the T2 being a distinct secure enclave is so that other processes cannot get to the data there. If they made it so there was "uniform memory access" between the T2 and the main memory ..... where is the security??????? if giving memory access you are letting x86 processes in to the data !!! How does that make it more secure? That's regressing.
Apple needs a better behaved and more secure interface between T-series and the rest of the system.... but hUMA isn't really it. At least for the majority of the T-series functionality. ( Apple may need more than 1-2 interfaces into the T-series and to internally compartmentalize it more. )
or something else. To control the whole memory space can be difficult (and therefore expensive) to maintain with old api’s.
Sign this seems to be more of a ARM takes over the Mac Pro heavy computation workload hand waving. Which the T-series isn't a "path " to that.
