Pure speculation, but I wouldn't be surprised if Apple weren't collaborating with AMD on GPU/chip designs. They seem pretty tight of late and AMD's forays into ARM powered servers were interesting but lacked the marketing clout that Apple could bring to the table. A scaled up ~20W A11X processor with Vega cores could be very compelling.
Doubt it.
As per my post above, any silicon dedicated to running X64 code is dead die space once the transition happens. Apple would be better off just throwing resources at making A12X (or whatever) run all native code faster and emulate the x64 using software. The space the x64 hardware would consume would be far better used by more ARM based processor cores, cache, more GPU cores etc.
That way they don't need to debug x64 (a complex architecture) in hardware - hardware debugging and fixing is expensive. They aren't wasting die space. The performance per watt on native code will be better, the cpu will be much cheaper to make, etc. And it will be a subtle push to developers to get their code ported, rather than just providing dead-end native hardware for it to run as well as native.
If Apple and AMD were in bed on GPUs, i doubt Apple would have designed their own. Maybe Apple might license infinity fabric or HBCC tech, but that's about it. If that.
I'm an AMD fanboy as much as the next guy (I have a crossfire vega 64 setup in my PC!), but it just doesn't make sense from Apple's perspective - AMD have nothing to offer them.
[doublepost=1527644888][/doublepost]
Would this run macOS or iOS qpps?
"Yes"...
[doublepost=1527645024][/doublepost]
to be honest, for myself and a few mac using buddies, the ability to run windows and similar is an absolute requirement and not negotiable. that being said I do not doubt for a moment that Apple would make this move seeing how much of a cut they get from the App Store and putting the Mac fully behind that walled garden would cement those profits.
and lose nearly every windows user and for games the steam players too ...
now if this is simply an education machine, destined for schools only it would be a shot across the bow for many low end manufacturers
Mac hardware is currently crap for gaming even if you boot-camp it. There's a massive pool of iOS game developers out there who would be chomping at the bit to get their software on some proper hardware - like a high end mac A series chip.
For the cost of a Mac that runs games at any real sort of pace, you could buy a 21.5" imac AND a gaming PC. I know this, because it's what I've done. Well, replace 21.5" iMac with Macbook Pro.
[doublepost=1527645355][/doublepost]
Burst speed is different than sustained speed. Apple used this method to justify AMD GPUs over Nvidia years ago.
The A10X is
passively cooled. In a very tight thermal enclosure... give it better cooling and the clocks will sustain much better.
I would not be so quick to make that statement in light of Intel's new 15W quad-cores which are currently available.
The A10X is running in a lot less than 15 watts. And was out almost 12 months ago. Time moves on. Apple won't be putting the A11X in a macbook. It will be their NEXT generation of part. With a larger die, more cores and improved tech.
[doublepost=1527645611][/doublepost]
What desert island have you been on? The latest intel chips have extra cores given 40-50% performance boost.
Did you read my post where i mentioned coffee lake is the FIRST UPGRADE WORTH TAKING since sandy bridge.
Of course i know they are getting potentially a 40% performance boost. When turbo-ing. At above the chips rated TDP, which can't be sustained.
That's the first time in 8 years they've made a jump like that, apple have been getting 50-100% improvement per year, since 2008.
Too little, too late - and their onboard GPUs are crap too.
Coffee lake is the exception rather than the rule - and is 5-6 years late.
[doublepost=1527645919][/doublepost]
As for Windows compatibility, Microsoft has been adding ARM support too. It’s where the world is moving. But even so, we’re moving into a post-Windows, post-Intel world. Apple’s plan is to leverage their iOS developer community to make Mac Apps, while providing backwards compatibility for some time. They’re probably working on making it stupid easy to port them.
Yup.
Even microsoft are trying their best to kill win32. With the new metro/fluent/whatever applications. Dabbling with ARM, etc.
The fact they've moved all their stuff to 365 and are now giving first class application support (almost, give it time) to iOS and macOS is yet another signal that they're going OS-agnostic.
Microsoft are just having a more difficult time than Apple will, because they've got 30 years of hacked-together-crap in Windows to sort out, and the applications that make use of it.
Apple went through a lot of the hard work in the shift to OS X already - macOS/OSX has had the benefit of a lot of lessons learned when NEXTSTEP was developed and also since. Plus apple has been much more brutal about deprecating and removing old technology they no longer want in the platform. Microsoft has not.
Those chickens are coming back to roost now. For everyone who whines every time Apple remove stuff from macOS -
this is why they do it. To enable platform shifts like this to be accomplished much more easily.
[doublepost=1527646045][/doublepost]
Did you mean Cannon Lake? That is the one that is supposed to finally bring 32G of mobile memory (LPDDR4) to the table. I’m withholding expectations until the i7 mobile variants actually ship. Those are the ones that Intel waits until the end of the year to indicate that it won’t be out in the next calendar year...leaving Apple’s options stuck at 16G for years now.
No, i mean coffee lake due to increased core count.
Cannon lake will be another worthwhile upgrade for Apple, however Apple has been able to use DDR4 if they wanted to make that call for years.
They didn't, and thats their call.
I'm talking worthwhile upgrades overall - not just from Apple's perspective. But even from Apple's perspective, the higher core count Coffee Lake mobile parts are a no brainer. It won't give them everything they want (no LPDDR4), but its actually more than a 5% improvement this time around.
But still an illustration of why Apple want to get off intel, long term. Apple's goals and Intel's goals don't line up well enough.
[doublepost=1527646230][/doublepost]
Because:
1) They're not "incredible chips" overall. They're mobile processors, and have performance compromises because they have to fit into an extremely low power and thermal envelope. They're incredible for mobile devices, but not high performance computing.
2) They're not compatible with existing Intel software, including Windows through virtualization. Running such software will require emulation which will slow things down even further.
Remove those power, thermal and cost constraints and Apple will be able to scale their A series chips up significantly. Just as intel scale DOWN their core series.
Double the core count and 1.5x the clock speed inside of 15 watts vs. the A10X (which would be totally feasible for a 13" macbook air replacement) and you've got a monster laptop CPU...
As to "not seeing the downsides". Of course i see the downsides. There WILL be short term pain in terms of performance. But if Apple can get enough of a boost in performance from AX, they can mitigate most of it on non-native software (as they did with the intel shift) and then the following 2-3 years make significant improvements in running non-native code, and also phase out the non-native code as well - getting even more performance back.
Yes, this transition is
not without pain. But apple have the pieces in place (LLVM bytecode in the store to recompile store apps natively, multi-arch binary support in macOS, rosetta technology, an amazing processor architecture) to alleviate a lot of it.
Long term, it definitely makes sense for Apple. And Apple tend to take the long view. They've got plenty of money in the bank to be able to afford to take a risk. If the new low end machine with ARM doesn't take off - so be it. But i think it will.
Will this happen at WWDC this year? I'm not sure. But i believe it
will happen inside 18 months.
[doublepost=1527647105][/doublepost]
Then there are some people who still say "Intel doesn't make what Apple wants"
Which brings me to my 2nd point... Apple is a fairly small customer of Intel. I know in the old days Intel would make a custom chip or two for Apple. But those days are long gone.
HP alone orders 3 times as many CPUs from Intel than Apple does. Then add Lenovo, Dell, Acer and others.
I'd say Apple represents about 7% of Intel's CPU sales.
So... is it any surprise that Intel doesn't bend over backwards to suit Apple?
Of course. All the more reason for Apple to do it themselves.
I dunno... this whole thing is funny to me. People are trying to paint Intel as evil and Apple will be better off without them.
No, just pointing out Apple's strategy. Intel can build processors for whoever they want. Doesn't mean it is in Apple's long term interests to use them.
But the bulk of Mac sales are high-powered laptops and desktops. It's gonna be a while before Apple abandons Intel for those types of machines.
I know there are some benchmarks that put iPad Pro over Macbook Pro.
However... real-world tests might have a different outcome. We'll have to see how well MacOS (and your favorite Mac apps!) run on a desktop ARM architecture.
What? Apple has sold far more laptops than desktops for about a decade.
The macbook air has been their volume seller since about 2010.
Take an A10X. Double or triple it in size (i.e., 12-18 cores). Give it actual active cooling with a fan and heatsink/heat-pipe.
You've got something that will destroy a "mainstream" desktop i7 on multi-threaded code inside of 35 watts. At far less cost to Apple.
Sure, it will perhaps take Apple a while to get the confidence to put that part out, but they could conceivably do it today. They have the tech. They have the motive. They have the money in the bank.
Apple as a company is bigger than intel... there's no reason they need to accept what intel make.
edit:
replaced A11X with A10X. got my model numbers confused. point remains.