While we're talking about 99.999% of things, what percentage of DESKTOP applications can usefully scale to that many cores?
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.
Kuo: Apple to Launch Several Macs With Arm-Based Processors in 2021, USB4 Support Coming to Macs in 2022
- Thread starter MacRumors
- Start date
- Sort by reaction score
While we're talking about 99.999% of things, what percentage of DESKTOP applications can usefully scale to that many cores?
Glad I switched back to windows. Apple is only shooting them selves in the foot with this. Oh well. Was different for awhile.
I don't have a choice if Apple moves macOS to ARM. I'm off the train whether I like it or not.
Was talking hypothetically. You have workloads scaling to thousands of cores already on GPU compute. It will happen in niche workloads like scientific modelling.
The Mac Pro isn't exactly an everyman machine.
[automerge]1585281121[/automerge]
That's a choice. You either adapt your workflow or adapt your platform. You're choosing to adapt your platform.
Not a lot, but think how many such applications you can run at once with that many cores!
The question was whether or not there's a credible ARM alternative to the HEDT chips from Intel or AMD. The answer to that question was that you could have a 100+ core ARM CPU. That's not a credible ARM alternative to the current state of the art.
So, the question remains -- in an ARM macOS world, what the heck happens to the Mac Pro? What is Apple's solution there? I can see how ARM in a low-end MacBook makes sense. I can almost see how you might rationalize an ARM based MacBook Pro. But beyond that it's a real head-scratcher to me.
I also agree with others' skepticism that Apple are willing to segment their macOS market into two, incompatible platforms and all the unavoidable complexity and consumer confusion that would necessarily arise from that situation.
Take the netlist for the top of the line intel processor. Hack the instruction decoder portion of the netlist to handle Arm instructions.
Now you have an ARM processor with essentially identical performance as the intel processor, but with fewer transistors.
Nothing magical about x86-64 that makes it more powerful than Arm.
This could probably sink Intel. These babies will be screamers! I wonder if they’ll play nice with Chrome and all the apps out there. This is a big move by Apple but it makes sense, however, this now puts me in a buy now or wait dilemma regarding a new Mac laptop.
Sure Apple could do it, but how does creating a bespoke cutting edge processor for one mega-niche model make financial sense?
They can leverage all that work to make processors for lower end products (and vice versa).
When I designed CPUs for AMD, for example, we designed first for the highest end product. Once we did that, it was easy to modify those designs for the middle and bottom of the market (reduce clock speed, shrink caches and buffers, reduce number of cores, etc., remove one pipeline per core) It’s an incremental cost, but the difference isn’t as big as you’d think.
I don't understand why so many people are gung-ho for this. Do you really want to deal with going through ANOTHER architecture switch? With your old software breaking AGAIN a few years down the line?
I mean, since 1999, we've had:
- MacOS classic to Mac OS X
- PowerPC to i386, then i386/x86_64
- Deprecation of i386 (x86_64 only)
Now you want them to do it YET AGAIN? Meanwhile I can run a twenty year old game on my Windows 10 gaming computer and it works perfectly.
Apple should stabilize their ABI for the long-haul instead of trying to reinvent themselves every few years. As much as I love MacOS this is the one thing that drives me crazy about Apple.
I mean, since 1999, we've had:
- MacOS classic to Mac OS X
- PowerPC to i386, then i386/x86_64
- Deprecation of i386 (x86_64 only)
Now you want them to do it YET AGAIN? Meanwhile I can run a twenty year old game on my Windows 10 gaming computer and it works perfectly.
Apple should stabilize their ABI for the long-haul instead of trying to reinvent themselves every few years. As much as I love MacOS this is the one thing that drives me crazy about Apple.
They are going ARM so it's easier to write apps across the board, hoping that iOS/tvOS/ipadOS fanatics will switch to Macs so they can use the same app on the desktop. Kind of like using Messages on MacOS. It's not a bad line of thinking, and the ARM processors will be faster and consume less power.
As for x86 support - I use a PC for gaming, so I really don't need bootcamp anymore. Until macs support gaming and get decent graphics, many people will still have one around. It sickens me, the GPU options in macs. For a $3200 iMac, I get a two year old AMD GPU that is instantly outdated. My three year old iMac can't run Fortnite for my kids but a $100 Nvidia card in a $600 PC runs it at 120fps. Weird. I would assume the ARM processors will also have integrated graphics like the mobile devices, which would be an improvement over the garbage Apple stuffs into their macs. Could also mean we don't get stuck on the current minus two years Intel upgrade. Does a single mac have the 9th gen Intel processor right now? I have a four month old Lenovo at work with a 10th gen i7. Yeah, I get it isn't a big improvement, but that's not the point. It was the same story back when it mattered.
As for x86 support - I use a PC for gaming, so I really don't need bootcamp anymore. Until macs support gaming and get decent graphics, many people will still have one around. It sickens me, the GPU options in macs. For a $3200 iMac, I get a two year old AMD GPU that is instantly outdated. My three year old iMac can't run Fortnite for my kids but a $100 Nvidia card in a $600 PC runs it at 120fps. Weird. I would assume the ARM processors will also have integrated graphics like the mobile devices, which would be an improvement over the garbage Apple stuffs into their macs. Could also mean we don't get stuck on the current minus two years Intel upgrade. Does a single mac have the 9th gen Intel processor right now? I have a four month old Lenovo at work with a 10th gen i7. Yeah, I get it isn't a big improvement, but that's not the point. It was the same story back when it mattered.
Will current third-party apps still work? The last thing I want to do is wait for ARM-compatible versions of Office and other programs I use. I can't remember how the PowerPC to x86 went, but I don't think a lot of the programs were compatible.
Are you assuming that they’ll be much slower than x86 cores? If so why?
I’d wait and see how it all goes before abandoning the platform. Perhaps there’ll be great translation, good enough for speedy virtualisation even, perhaps the industry as a whole will nice to ARM app the docker images you need will be ported anyway.
I actually think it’s perfectly plausible they’ll allow a choice of chip architectures in the Mac Pro. It is modular after all.
It doesn't matter. They had an emulator called Rosetta for when they switched from PowerPC to Intel in 2006, but they killed it off only a few years later. Every time they do one of these architecture shifts they end up leaving legacy software behind. There's a lot of great software out there that's no longer maintained, and we always lose out when it gets left behind.
USB 3 isn't a connector. The connector you're referring to is USB Type-C. USB4 and all foreseeable versions of USB will likely be using that. I mean, at some point everything will go wireless I'd assume, so USB-C may be the last physical connector we ever need. It could certainly last a good long time regardless.
If only USB-C had a mini variant for its plug, probably make it more resilient to premature wear.
I could see the transition away from Intel to custom in-house chips being good or bad for Mac users over the long run. I'd worry about performance stagnating down the line if the company receives limited returns on investment in research. On the other hand, they'll almost immediately become a major competitor to Intel/AMD and possibly push performance forward across the whole market.
The main development I would hope to see from this transition is the Mac moving into innovative form factors. Perhaps that is part of what's motivating the transition? Not that current Macs are outdated, I'd just like to see a product line that continued to grow and evolve over time.
The main development I would hope to see from this transition is the Mac moving into innovative form factors. Perhaps that is part of what's motivating the transition? Not that current Macs are outdated, I'd just like to see a product line that continued to grow and evolve over time.
Wireless is still inefficient though there are ways to make it comparable to a wired connector the question remains will it be similar to USB-C an all for one connector.
[automerge]1585283529[/automerge]
Office 365 and Office for iOS, I don’t see this as a problem. Developers have been taking they time to port to ARM, Adobe seemed to have gotten the memo later compared to most companies.
[automerge]1585283722[/automerge]
That’s a problem of the developer, if not interested to port it then make the code OpenSource so the community can do so. Seems more selfish by the developer to just allow they software to die rather than donate it so it’s customers are not left out in the cold.
Last edited:
oh, c’mon. You sound like an anti-Apple troll. Or you’re quoting one. Most iPhone users are not Mac users. Just do the math. Pretty sure Apple sells more iPhones in one quarter than there are Macs in use total.
and lots of Android users ARE Mac users. Processor is so far abstracted from everything that it makes almost no difference. Did you know that Microsoft Office runs emulated on Surface tablets? So there’s that. And ostensibly ARM variants in Macs would be faster. So there’s that. And you know Apple would be running macos apps as hybrids.
And that's why these chips were purposefully designed for servers. Apple would not have any use for such chips even if they had high multi core Geekbench score
In the current Apple chips, their big cores are probably almost as big (in silicon) as x86 cores. The high-core ARM server chips mentioned above are narrowly focused on certain types of loads. For example, some people speculated that they have rather poor floating purpose performance (simple design saves area). One should not expect that Apple will be able to squeeze way more cores into their ARM CPUs than x86 vendors do. Besides, both AMD and Intel are going with multi-die designs these days (AMD being ahead in this department). Apple has nothing of sorts in their portfolio (nor did they need it until now).
Last edited: