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.
ARM v9 Instruction Set Officially Announced.
- Thread starter Colstan
- Start date
- Sort by reaction score
I don't need x86 architecture, but I don't want to buy an M1 MacBook today and suddenly got obsoleted in a few years due to ARM v9 transition.
That's not something I would worry about. Apple just introduced the M1 and they aren't going to obsolete it quickly, if ever. It's not like the upcoming V9 instruction set was a surprise to Apple. Since Apple makes their own processors, they may or may not support the V9 instructions. But either way, your M1 Mac will not stop working.
Refer to the first intel Macs with Core Duo chips. They got obsoleted very quickly due to having 32bit chip. Unlike Windows, when Apple wants to transition to something, they are not afraid dumping their own hardware from support very quickly. I'm just wondering if ARM v9 will force a significant architecture difference that Apple might drop support for ARM v8 (just like Apple dropping 32bit A chips).
V9 is a very minor set of additions. What are you worried about?
I think M1 will get outclassed in it's feature set (like external display support etc.) fairly quickly, after all, it's a first-get product with very conservative design, but M1 machines will continue to be supported and perform well for a while.
No, it doesn't. As @cmaier above writes, ARM v9 just adds some new features on top of the existing Aarch64. Most of these features do not even affect the regular application code (the security stuff, if I understand it correctly, is mostly a feature for the OS and the VM-software). The only really significant addition is SVE/SVE2, which already existed as an optional extension in ARM v8, which does have the potential to improve performance of some application and uses new instructions. We'll have to wait and see whether Apple will introduce SVE in their upcoming prosumer machines and how they will handle the feature set disparity.
Considering the scale of the roll-out of the M1 processor, it seems unlikely that Apple would abandon it anytime soon. Laptops, mini, iMac, iPad Pro, the chip is in quite a few places in the lineup now, and they reportedly made 20m of them for launch, so it seems unlikely it will be phased out soon.
Apple is Apple. They didn't mind cutting off 32bit clean, leaving some devices with quite short support cycle (eg. The core duo Macs). I would be very glad if my fears are unfounded.
Glad if that's the case. I have very minimal knowledge, but I read that v9 will affect security, thus it seems that it might require significant changes in the OS. I don't know. It's just that Apple's track record in their transitions was not good for the early devices. Hopefully I'm wrong. Would be glad if v9 is just an addition to current ISA.
V9 is just a few new extensions. It’s really not a big deal. The security stuff is an add-on, and it will be a long time before apple requires it. And the other stuff more or less improves performance in some cases. Remember that Xcode makes it very easy to bundle multiple binaries in a package, so there will be no problem supporting both architectures for the foreseeable future. This is a giant nothingburger.
Says the people that didn't think it was bad for x86/64 compatibility to go away. Apple has a really horrible idea of backwards compatibility, so I wouldn't bet on Apple ever remaining compatible generation to generation.
So, how long should an OS vendor support legacy software? How long should an organization continue to use and expect OS support for abandonware? At what point, or by which criteria, is it acceptable to stop allocating resources to support older software and shift them toward new technologies?
I find Apple somewhere in the middle between Google constantly chasing after the new shiny, and Microsoft bringing along a load of cruft with every release. Given how quickly Apple's customer base tends to adopt new releases and developers have released updates, this doesn't seem like a widespread issue. With the sales of M1, customers seem pretty happy overall, even given some of the hiccups.
I find Apple somewhere in the middle between Google constantly chasing after the new shiny, and Microsoft bringing along a load of cruft with every release. Given how quickly Apple's customer base tends to adopt new releases and developers have released updates, this doesn't seem like a widespread issue. With the sales of M1, customers seem pretty happy overall, even given some of the hiccups.
Last edited:
It wasn’t bad for x86/64 compatibility to go away (Nor Has it actually gone away).
This is not a new architecture. It’s some new instructions. It happened all the time in x86-land, and nobody every got all crazy about it. Calling it “v9” was generous. It’s marketing. It’s no different than when Intel or AMD adds a few new instructions from one chip to the next. We did that all the time at AMD.
It hasn't gone away, for now, as long as you have an Intel Mac, otherwise it's already gone. And it will be gone in a couple of years for everyone, that's Apple's typical MO.
Like I said, I wouldn't bet on it. Past actions speak volumes.
How many people?
Should macOS have 6502, 68K, 32 and 64-bit PPC, and 32 and 64-bit Intel compatibility layers with corresponding OS support? Even MS has ditched 16-bit support.
I’m still waiting for intel and/or amd to ditch 32-bit (and lower) compatibility. It would make the processors much simpler, somewhat faster and more power efficient. Who cares about 32-bit now?
Any.
In Windows, I can run 16-bit DOS in a VM if I want to (and do actually). I don't care if that legacy is emulated or in a VM, it's just got to be there, Apple doesn't have that. I can't run an older version of MacOS on my M1 Mac, nor can I run my x86/64 VM's.
I doubt if they will, especially without any way to emulate it. They understand legacy support is important to the people that buy their machines. After all those machines run software, and that is why people buy them. People that don't care if what they buy is simpler, and the power savings you are talking about isn't enough to care about.
There an awful lot of people who run *no* 32-bit software, and the price those people pay (in performance, power consumption/dissipation, and even side-channel vulnerabilities) due to support for these unused features is very high.
So? Does every processor sold have to be suitable for every possible use?
If third parties want to offer support for older software, I say go for it. But I don't believe that Apple should be obligated to do so. Unless MS decides not to license WoA for VMs on Apple Silicon, it looks like there will be a path to running Windows software on the Mac, including x64 emulation. Perhaps SheepShaver will make it to M1.
From a business perspective, eventually the costs of legacy support outweigh the benefits.
Change happens constantly in the universe, even if it is on a scale we can't always readily perceive. You can view change as a hassle, or you can view it as an opportunity for improvement and growth.
From a business perspective, eventually the costs of legacy support outweigh the benefits.
Change happens constantly in the universe, even if it is on a scale we can't always readily perceive. You can view change as a hassle, or you can view it as an opportunity for improvement and growth.
Of course not. I really don't care about the hardware side at all, I just have software I need to run, and have been able to run it on Intel Mac's. I can't anymore, so that effects me and is fair game for complaint.
That's where you are wrong, the ROI on rewriting software for a new platform is never positive. It's a money flush down the drain, and for no benefit. I'm the corporate guy that buys the hardware and spends on the software, it's what I do for a living, and I have never seen a good money case for a straight across rewrite, it just doesn't make sense in anything but the computer business itself. Where I work, computers/software are purely an expense...