Mark Gurman, Bloomberg:
Steve Jobs long believed Apple should own the technologies inside its products rather than rely on mashups of components from other chip makers, including Samsung, Intel and Imagination Technologies. In 2008, the company made a small but significant step in that direction by acquiring boutique chip maker P.A. Semi. Two years later, Jobs unveiled the iPad. The world focused on the tablet’s giant touchscreen, book-reading prowess and creativity apps. But the most ground-breaking technology was hidden away inside: the A4, Apple’s first processor designed in-house.
And:
Apple’s push into the complicated and pricey chips business makes sense so long as the company is selling 300 million devices a year.
An interesting read. Lots of solid reasons for Apple to roll their own chips. Cost savings, for sure, but there’s also self-reliance. They can control chip production and act quickly if and when a flaw/vulnerability is discovered.
But most importantly:
So far, only two Mac lines include custom Apple processors: the MacBook Pro with Touch Bar and the iMac Pro. Apple is working on at least three updated Mac models with custom co-processors for release as soon as this year, including updated laptops and a new desktop, according to a person familiar with the plan.
Well, that’s pretty big news. And a logical next step, if true.
Will we see a Mac Pro this year based on Apple silicon? How about a MacBook Pro? Can’t wait.
The only way the development cost of the T2 chip makes sense is if Apple rolls it out into every single Mac they make.
This is going to make a lot of companies that sell controller chips very unhappy.
More rumors, dressed up as news. We already have MacRumors for that, Dave.
“Will we see a Mac Pro this year based on Apple silicon?”
(invokes Betteridge’s Law)
ARM chips aren’t powerful enough to run the Mac Pro. On the consumer models and MBP… the problem isn’t the OS and not really performance from the looks of it… it’s that Mac software is all built to run on x86 CPUs and would need to be ported. Given the relative stagnance of the desktop/laptop segment… would developers do that or would many simply bail? They’d need most of the apps that most people run (say, the top 1-500) to be available day 1 – there’s no way I’d buy a machine running all my apps in emulation.
And then… what’s the selling point to the consumer? Apple could do this for their own reasons eventually, but why would I want this? One reason would be substantially longer battery life. Aside from that… I can’t think of reason I’d care about the CPU.
It’s logical that the T-series chips will eventually be in all Macs, including the new Mac Pro. But using A-series CPUs in the Mac is an entirely differently thing.
I know the A11 does very well in benchmarks vs mobile Intel chips, but that alone doesn’t make it a realistic replacement for Intel chips across the Mac line. For example: I don’t think it supports Thunderbolt and PCI, 128GB of RAM, ECC RAM, etc. I’d imagine that adding that stuff is non-trivial, and maybe not financially prudent given the relatively small number of Macs Apple sells.
Switching CPU architectures requires enormous effort on the part of Apple and it’s developers; and several years of adjustment for it’s users. They did a great job with both previous switches, but it was still a pita. Not sure the necessity and payoff is there this time.
Coprocessors ain’t gonna make it “based on Apple silicon”, so, no, not this year.
(I’m also not really sure Apple would be any faster about fixing bugs in big silicon than Intel would, because the problem is that the task is really difficult, not that Intel [or ARM] is especially bad at it or Apple is especially good.)
Ported?
Recompiled, unless your code is so heavy it actually has embedded assembler.
(Or other weird, uncommon stuff.
Adobe and Apple might have to do some work; I don’t think 99% of developers and solutions would need more than a recompile and smoke test, honestly.)
Looking at iFixit’s T2 description from the iMac Pro teardown [and what the iMac Pro has on the MB], I don’t think the “controller chip” vendors are terrified.
Since it still has a separate audio chip*, several Intel controllers, separate Ethernet and Wireless controllers, PCI mux/demux, memory controller …
(* No idea what the T2’s “audio controller” does vs. the Cirrus audio/DAC chip; maybe Apple will clarify.
Eventually Apple might well in-house all of that, though I’m not sure what their incentive is.
But the T2 ain’t it.)
Ideally, yes. But things rarely go that well. Even if they do, devs now need to do two versions of an app, the x86 and ARM version. They need to test both, etc. And that assumes all of the APIs are the same. If not, then….
Both good points. Intel’s game is silicon, so they have a lot of incentive to avoid bugs, and fix them in one iteration. It’s not at all like Android upgrade releases, for example.