It's funny how other OEMs are able to build high-performance and low battery life laptops using x86, but Apple struggles. I wonder why that is?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
If Mac went ARM tomorrow, would you jump?
- Thread starter russofris
- Start date
- Sort by reaction score
Who would want a low battery life option?
Well, Apple's notebooks are not bad in terms of battery life (they've been better in that regard in years prior though). Maybe Intels have a good power/watt ratio. However, performance per watt is where ARM is actually pretty good, so why not take advantage of that in their mobile computing line-up?
I'm not saying Apple will or should transition. But why not consider all options on the table? In particular ARM has come a long way; its the first time in a long shot that there even is an option other than x86. And who knows what Apple is cooking up in their super secret labs?
Interesting times, anyhow ...
Why not stick MacOS on an iPad Pro... (and release a "smart keyboard" with a trackpad)?
I would be very surprised if they haven't got a version of MacOS and some of their own applications running on ARM (likewise they had it running in x86 before they announced that switch) although whether its 'stable' enough to let out of the building is another matter. Trouble with "fooling around" is that they'd need a critical mass of applications to make it worthwhile, which means getting third parties in on the act. Its not going to be zero work porting anything to ARM (even in the 'check the ARM box and click Build' scenario the result will have to be properly tested) so you'd need a bit more justification than ;just fooling around'.
Who would want a low battery life option?
Do laptop users care about power/watt after a certain point? I'm sure most consumer laptops rarely leave the house. For those that do, there are Windows laptops that get 8+ hours of battery. Would people who need portability really want to give up compatibility for the sake of 16+ hour battery life? It seems like a solution in search of a problem, much like the MacBook Pro's touchbar, removal of ports, etc.Who would want a low battery life option?
Well, Apple's notebooks are not bad in terms of battery life (they've been better in that regard in years prior though). Maybe Intels have a good power/watt ratio. However, performance per watt is where ARM is actually pretty good, so why not take advantage of that in their mobile computing line-up?
I'm not saying Apple will or should transition. But why not consider all options on the table? In particular ARM has come a long way; its the first time in a long shot that there even is an option other than x86. And who knows what Apple is cooking up in their super secret labs?
Interesting times, anyhow ...
Yes unfortunately AMD only really has reasonable mobile chips in the 15W TDP range (not the 35W and 45W TDP laptop range used in the 15 inch MBP where Intel still is king) and even then it would be a side-grade (slightly worse CPU performance for better iGPU compared to chips already in the 13 inch MBP).
AMDs current chips would only be a nice improvement for desktop macs like the iMac (which just got refreshed anyway) and an 'xMac' that is unlikely to ever exist, plus Apple likely get discounts for going all Intel as you say.
If they ever go all-ARM on desktop I will likely wait many years before considering Apple again (if ever) and just stick to PC hardware. The bulk of macbook users (who are responsible for most Mac sales) probably would not care either way about ARM vs x86
MacOS on Intel isn't going away overnight.
...but then, with the current train-wreck of desktop offerings (...a Mini that needs an external GPU just to smoothly drive a pair of 4ks, an iMac that comes with a like-it-or-not display welded in or a Swiss Cheese Grater offering off-the-peg at bespoke prices) a lot of people are switching to PC anyway. Sadly, I don't think Apple are ever going to make an "xMac" with any processor - be it Intel, AMD or ARM - and the Hackintosh option will go away once all< 3 year old Macs have T2 or later chips.
At least a switch to ARM might mean cooler, quieter Mac Minis and iMacs with more proportionate graphics (if you really must make Mini or super-thin all-in-one then you really want a desktop-class CPU with premium integrated graphics - a combination that Intel just don't offer).
Personally, I'd love an ARM desktop - if only for nostalgia's sake - although the new Raspberry Pi 4 is sounding good and seems to have fixed the horrible i/o bottleneck that knobbled the previous versions - and you can get 171 of them for the price of a Mac Pro
Still, I'd prefer something that could take PCIe, M.2 cards and SATA... but Apple ain't gonna make that either.MacOS on Intel isn't going away overnight.
...but then, with the current train-wreck of desktop offerings (...a Mini that needs an external GPU just to smoothly drive a pair of 4ks, an iMac that comes with a like-it-or-not display welded in or a Swiss Cheese Grater offering off-the-peg at bespoke prices) a lot of people are switching to PC anyway. Sadly, I don't think Apple are ever going to make an "xMac" with any processor - be it Intel, AMD or ARM - and the Hackintosh option will go away once all< 3 year old Macs have T2 or later chips.
At least a switch to ARM might mean cooler, quieter Mac Minis and iMacs with more proportionate graphics (if you really must make Mini or super-thin all-in-one then you really want a desktop-class CPU with premium integrated graphics - a combination that Intel just don't offer).
Personally, I'd love an ARM desktop - if only for nostalgia's sake - although the new Raspberry Pi 4 is sounding good and seems to have fixed the horrible i/o bottleneck that knobbled the previous versions - and you can get 171 of them for the price of a Mac Pro
I don't understand. What's to be gained with an ARM-based desktop, especially against an AMD-based desktop? I'm just trying to learn something new here.
...I did say that, from my POV, it was partly nostalgia (I used ARM-based Acorn machines for years back in the day). However, it would be good to have more than one personal computer processor architecture on the market.
The issue with Intel and AMD each only manufacture a fixed range of processors and systems-on-a-chip with whatever combinations of CPU, GPU, cache, I/O, acceleration hardware , power consumption etc. they deem fit. On their own timescale. Intel have a nasty habit of introducing, with great fanfare, their new generation of "whatever lake" processors while Apple is still waiting them to release the previous generation 36W-quad-core-with-Iris-pro-plus-graphics chip that they need to upgrade particular Mac models.
ARM, on the other hand, don't acually make processors - they just license their designs at various levels from just the instruction set, though processor core and GPU designs up to complete system-on-a-chip designs. A company with Apple's resources might not want to design a processor from scratch, but they can build the exact Ax-series chip that they need by licensing the bits from ARM and other third parties.
I already gave the example of the Mac Mini, where Intel just don't make desktop-class chips with more than the most basic integrated GPUs ( a: Desktop PCs usually have at least one PCIe slot and b: Windows doesn't rely on GPU-heavy non-integer resampling to drive 4k displays at variable resolutions). Apple could make an A-series chip with a desktop-class processor and a much better integrated GPU for the Mini and lower-end iMacs.
Also, generally, each ARM processor core is a lot simpler than an x86 core (modern x86 processors actually consist of a RISC-like core plus hardware to translate x86 code into RISC operations) so you can simply fit more on a chip - or have space for more specialist graphics/math/media codec/encryption hardware on the same chip. There aren't many "desktop" ARM chips out there at the moment, but there are some ARM server chips with 32 cores that show what's possible.
Oh, this is informative! Now I have more questions.
I often wonder why Dell, HP, etc can make very good Intel laptops, yet Apple seems to always be "waiting for" or "disappointed in" Intel. Could you please help me understand why that is?
Also, wouldn't Intel NUCs and other mini-PCs be superior alternatives (OS aside) to the Mac mini?
Finally, I've always wondered what's so great about integrated GPUs, ARM or otherwise? Can they really compete with discreet GPUs from AMD and Nvidia?
Dell and HP are stuck with the same Intel silicon as Apple. They simply make different compromises. You can have 32GB of memory but then you can't use LRDIMM's and memory power usage is way up, so then you either need a honking great battery or battery lifetime is short. You can stick a discrete GPU in, but then you're sucking WAY more power and space. Apple's retina displays are gorgeous but are harder for a GPU to drive, so the crappy end of an integrated GPU line probably doesn't work with acceptable performance. Apple wants a laptop with particular form factor, weight, battery life, and performance requirements, and you simply can't tick all the boxes simultaneously with the Intel CPU's available today.
I won't get into the mini question since I've not looked at the Mac Mini for many years, but I suspect that the answer is similar: it depends on what compromises you want to make.
As for integrated vs discrete GPU's, the integrated GPU will never come anywhere close to matching the high end of discrete GPU's. You need very high speed memory access, with different access patterns than a normal CPU has, and you need impressive bandwidth to the screen. I don't expect to see all that on-chip for years if ever. What is nice about an integrated GPU is that it's smaller (duh!) and uses substantially less power than a discrete GPU. Unfortunately, Nvidia has been the performance-per-watt leader in GPU's for some time now, and Apple and Nvidia don't seem to get along.
Honestly, intel’s Staggered release schedules since Kaby Lake has to be driving all PC vendors a bit nuts.
This all started with Broadwell, though, and it’s been f***ed up ever since. Still waiting for Intel to catch up on 14nm and actually deliver 10nm in volume. I still don’t believe they will in time for the holiday shopping season, but I digress.
Pretty much what @kschendel said - some of other companies' upgrades have been of questionable value if you look carefully at the specs - they may have downgraded from a 'premium' iGPU to a regular one, or switched to a lower-powered (i.e. slower) processor model just to get the "New 9th Gen Processor" label. This is aided and abetted by Intel's "i3/i5/i7" branding which is almost pure marketing and tells you very little about the specs. Also, since all of Apple's Mac products are ultra-slim or small-form-factor their designs tend to be very dependent on the performance and thermal/power characteristics of a particular range of processor models.
...also, the MacOS GUI is quite GPU intensive (esp. with non-integer scaling, which is the default in some models) so it can't really make do with a sub-par iGPU.
On the other hand... Intel's release cycle is neither a new phenomena or a closely guarded secret, and Apple have huge resources. They could probably try a bit harder - not wait 4-5 years between major re-designs, maybe diversify their Mac range to include some models that prioritised power over thinness. I can't see any reason - for example - why the 2018 Mac Mini got 8th-gen chips 6 months before the iMac. I suspect time/money investment in Mac takes a back seat to the more profitable iPhone and that Apple are heavily relying on MacOS lock-in to keep the Mac customers on board.
Many of the NUCs and similar machines are a bit "fake" in that they look very neat in the photos - where you can't see the humungous external power brick or hear the fan going insane. The Mac Mini is almost unique in having a built-in PSU.
However, yeah, on the face of it, Apple skipped an opportunity to build a "Skull Canyon" Mac Mini, which would have been a major boost c.f. the 2014 model, and the Hades Canyon - with semi-discrete AMD graphics (i.e. there's a CPU and separate GPU chip combined into a single physical package) would have seemed like the obvious choice for the 2018 Mac Mini.
If the OS was not an issue, if I was looking for a Mini PC I'd certainly consider either the NUCs or one of the many Mini-ITX or smaller PCs that offer a more sensible balance of size and expandability than the Mini. Plus, all joking apart, if you're happy with Linux and don't need high performance check out the new Raspberry Pi (about $70 by the time you add a case, PSU and SD card).
I hate to say it, but the 2018 Mini is a blatant example of cutting the cost while increasing the price and convincing the faithful that it is somehow an "upgrade". However you try to rationalise it, a quad-core desktop-class i3 is a substantially cheaper component than the contemporary mobile-class i5 processor that would be the natural progression from the older models.
Cost, space - it's one less chip on the motherboard - and power consumption/heat. The Intel ones, certainly, can't compete with discrete AMD/NVIDIA GPUs but the better ones are good enough for general use. They're certainly the sensible choice for something like a MacBook Air.
...but MacOS really needs the better Iris/Iris Pro (or whatever Intel is calling them today) integrated GPUs - which Intel only builds into laptop chips. The iGPU in Intel desktop chips is only really fit for basic "business graphics" - not running multiple 4k or 5k displays in GPU-intensive scaled mode.
It will be interesting to see if AMD's new RDNA graphics processing building block shows up in a Zen 2 laptop TDP processor down the road. RDNA ought to give the Ryzen APU (iGPU) the necessary performance at a lower power draw than the current Vega-based APU's.
I would love to see Ryzen in an Apple product, especially if AMD can keep the momentum going. The 2700X is a sweet multi-tasker and the 3000 series should be even better.
I would love to see Ryzen in an Apple product, especially if AMD can keep the momentum going. The 2700X is a sweet multi-tasker and the 3000 series should be even better.
Okay, one more question...
How would virtualization work on an ARM-based Ax? Both Intel and AMD's x86 processors support hardware-assisted virtualization. Would Apple even care about something like that?
How would virtualization work on an ARM-based Ax? Both Intel and AMD's x86 processors support hardware-assisted virtualization. Would Apple even care about something like that?
I don't think we'll know anything until or if Apple announces an ARM chip. That is assuming they are open about the available feature set.
I think the most likely short term scenario for this will be a Mac with both an ARM and an Intel processor. Mac OS, as well as all native apps will run on ARM, freeing the Intel CPU up entirely for your third party applications. When using only native apps, the Intel CPU can be put to sleep, giving you better battery life.
This is just magical thinking. We certainly know enough about ARM to know that a theoretical ARM CPU with hardware-accelerated amd64/x86 virtualization is just not a realistic expectation.
I mentioned virtualization. I didn't say anything about amd64/x86 virtualization. There is nothing stopping Apple from including virtualization for ARM in their chip... well, except Apple.
Sure, but that's a market with zero customers. When macOS users express concern about losing virtualization, they're talking about their need/desire to run VMware, Parallels, and Docker for Mac (among other tools). Nobody gives really two blinks about ARM virtualization on ARM.
Sure, anyone currently using the x86 virtualization would express concerns about losing x86 support. That is why I wouldn't buy an ARM Mac. That doesn't mean you can't have ARM virtualization. If I did buy one, I would definitely care about ARM virtualization and what other operating systems I could run.
Regardless, I read it as a question about generic virtualization because of the features of the x86, not about x86 specific virtualization. A market for a non-existent product is probably pretty hard to determine at this point. We'll have to if or when Apple goes in that direction.
When I say market I meant the greater ecosystem of amd64/x86 virtualization container formats. There is a thriving marketplace of docker containers, Amazon AMI images, VMware OVF templates, etc. Platforms such as Kubernetes and Docker Swarm. Cloud hosting services on Amazon and Google and others. These are all foundationally built upon amd64/x86 virtualization. macOS is a first-class citizen when it comes to development, testing, running, and deploying to these technologies.
Compare that to the equivalent ecosystem of ARM container formats and platforms, which is basically non-existent besides a handful of nascent and fragmented IoT and edge layer stuff like K3S on ARM. The emergence of ARM macOS virtualization capabilities would not really impact, influence, or legitimize this technology in any appreciable way. If macOS ARM did see a port of, say, virtualbox for ARM virtualization, you'd be able to run... I dunno... ARM Alpine Linux in all its glory. That's better than nothing, but it isn't even in the same league as what can be done in macOS today on an Intel platform. Not even worth comparing.
All of this completely ignoring the even more widespread desire to run virtualized Windows for Windows application compatibility and we can definitely say that is not going to be possible on virtualized ARM.
You would probably be able to run most flavors of Linux and possibly Windows 10. The market is non-existent because the world is populated by x86 and there has been no widespread need for ARM virtualization. The chip market has the corpses of many that failed to successfully compete against x86.
I don't see Apple wasting time on it but they could. I'm not advocating for it. We were just talking about two different things.