Apple switch to use ARM cpus, good or bad for Cubase?

It seems like the bumpy road started for Mac users with Mojave just get more bumpier. Will it be good or bad for Cubase?

Hi,

No one knows at this moment, I would say.

Absolutely no way this could possibly be good. Even if Rosetta 2 works for pro audio apps, even if somehow magically Rosetta 2 doesn’t impose a performance penalty, even if somehow Apple finds a way to make computers with improved performance, this will be a huge distraction and expense for developers. It’s a zero sum game. If Steinberg must spend time and money supporting Arm processors (testing alone will be expensive), that’s time and money that isn’t spent on new features. In the end, there will be apps and plugins that are left behind.

This change will improve profitability for Apple. It’s unlikely anybody else will benefit from it.

For sure it will be costly for the years where you need to support both. Steinberg also share a lot of code with windows-intel so much of optimization are likely to be the same or at least not collide. But there also other things will apple even allow ASIO for ARM even if it would be technicaly possible. Do they use ASIO for their cubasis ?

Microsoft Also uses ARM in Some devices from 2018
I expect that in the future ARM will be the strongest and best
In terms of processor temperature, thickness, power, and energy consumption
There are ARM processors now with a power of 160 cores dual socket Ampere® Altra®, Altra® Max Cloud Native Arm Server Processors

I expect programmers around the world will support this CPU

1 Like

Won’t be an issue, Steinberg moved across PowerPC CPU’s and Intel… And frameworks are so much better nowadays that software sits on top of, people who’re working on that level you’d expect would be totally separate to those on the detailed interface and functionality work.

More concerned about Mac users who’ve invested big bucks into Mac Pro style items who may be forced into running new updates to their apps within the Rosetta 2 emulation layer. This is why i’ve moved to a windows system, can just watch from afar and not get caught up in the headaches.

As for if it’s good or bad, my understanding is that ARM still has some way to go before it’s got Intel performance. And i haven’t a clue how single core performance will be in real use, which is really important for single threaded functions that DAWs utilise. Give it a few years and if it improves the power consumption and increases productivity then great.

Intel chips nowadays really rely on the turbo boost to hit full performance, i don’t know whether the ratio is so high with ARM, or if they will be more like the older Xeon CPUs whereby you had the single core performance 24/7 without reliance on turbo boosting.

But, i don’t know… I really haven’t a clue what the equivalent cost of an i7/i9 machine would be in ARM world. Will it be cheaper? Will it make the new Mac Pro look cheap?! The only certainty is that software ‘will’ be transferable, no-one would be mad enough to obstruct that… But if it’s unstable, wow it’s gonna get messy!

I also a sneaky feeling i have that Linux may rise in the ranks with the move to ARM. Change in CPU brings uncertainty, and linux offers a cheaper, robust alternative for those who want to ‘wait it out’ on a secure system.

Intel has 224 core hand have had for a view years with server XEON’s. (8 socks and 1.6 kW)
but they dont perform that well in SMP. AMD has better SMP with their dual EPYC (64 core each)
The Altra does not state that is 160 core is a “mesh-based interconnect topology” so it wont fit for
workloads like Cubase and most DAW’s that needs a SMP, you can maybe do something 2 or 3 zone NUMA
but it wont be very well balanced.

Thanks for your explanation
I was hoping for a good future for these ARM CPUs
Because the processor heat
Experienced by Intel and AMD

Personally last year I switched my hexacore 2018 Mac mini to Windows 10… And I have absolutely no regrets, and I’ll - never - come back to macOS!

What upsetted me the most was when Apple stopped developping Rosetta after Snow Leopard: At this time, I had some old plugins / apps that I really really cared about and used daily, and many of them they was - never - updated to the “new and shiny” Intel architecture!

It’s like Apple don’t care at all about the - COSTS (I talk also about human cost) - of development / transposition to new frameworks / IDEs / languages… They say it’s easy! It’s work out of the box!

…But the reality of PRO / specialized apps is: it’s difficult, it takes a LOT of times to obtain the same performance / optimization with the new system, the reality is that, for a lot of biznesses / entrepreneurs which have bet on Apple sh*t: it’s the end, we close, bye bye! /:

Think about this simple fact: With Win10, I can launch and really use apps as old as… Win 95!

For many people and biznesses, compatibility is absolutely, essentially, existentially… primordial! You can’t just jump on a wagon like that, you have to think deeply about the consequences, about your projects, about your clients and your employees, and take very, very serious decisions that engage the future of your group and, if you think about it, your life too.

They might very well be very good for some work loads. But the workload a DAW has is more or less it’s worst case.

The iOS pricing structure will become a real problem for plugin and DAW developers on macOS because on the long run you’ll be able to use iOS music apps and plugins on the desktop too. That sounds great at first but why would any sane user spend $150 for a macOS plugin if you can get an almost identical iOS version for $30 (eg Fabfilter, Waldorf, etc)?

Same for DAWs. If they are good enough to create music on a tablet then they’ll work equally well on the desktop. Many users will even be upset if they were told to buy the same software again (eg Cubasis) for the desktop for a lot more money. Is the “real” version ten times better than the mobile version as the price suggests? Not for everyone, I’d guess.

If they merge it will be a problem. If they keep OSX “open” (eg root shell) that is. It is not far fetched that they will use IOS on Mac too. Then they are the same. Everything need to go though the app shop, no need for dongles. And you can not have the same license for Windows and Mac. But we are not there yet, it might be the opposite too that you could run OSX on a ipad. (It less likely but for developers it would be nice) And for developers it is a big Apple TAX they need to pay for selling the stuff on App Store. But we are not there yet!

Yup i think MacOS will go closed like IOS. Because that means that Apple can shaft every Mac owner for more iCloud subs as your primary ‘sharable’ storage. their cards have been on the table for a long long time now. Hope i’m wrong, but it also wouldn’t surprise me if they will also do their damnest to lock google cloud services out for those that dare want to be crossplatform.

This is why i can see Linux rising significantly in the coming years.

That’s definitely the way the wind is blowing. The end game is: apps only come from Apple, supplemented with some trivial curated utility apps from the app store. The switch to ARM will cull some 3rd party apps. Sandboxing and security requirements will cull some more. The race to the bottom for app prices as the distinction between MacOS and IOS is blurred will drive even more developers away.

Apple has been openly hostile to pro users for some time now and this is the next step in cutting that segment of the market loose. I believe they’ve made a strategic decision that they want to serve the 90% of users who aren’t very demanding. The other 10% can go elsewhere. That’s a reversal of the Jobs era where pro users were courted since they were seen as influencers. Now pro users are seen as annoyance.

So where do you put Cubase in this? Does it it belongs to the pro or the other, should we go elsewhere?


No one except Apple knows yet how well their chips scale for desktop use right now (I think it’s reasonable to expect: quite well), but for laptops this will be an unequivocal win. The transition period has its own difficulty, but it’s absolutely worth it long term, just like the switch from PPC was.

Intel has not managed to substantially improve their performance per watt and now they’re being flanked by both Apple and AMD. Intel really dropped the ball while having a near-monopoly for a while. Apparently they thought their win was so complete they could just half-ass their way into the future with some minor updates. Really can’t blame Apple for ditching Intel after the PR embarassments caused by Intel’s thermals for them.

And just overall I’m extremely happy to see we’re back to a more healthy market situation after Intel dominating for so many years (as a sidenote, I’m quite happy to have an AMD CPU again, finally, because it’s the best for me, not just because I want to encourage competition!)

That’s not the question being asked here though, it’s what Apple uses as their focus for evaluating a successful chip/performance development with their OS that matters to us Audio peeps.

Yes, for 90% of users the move to ARM could be superb, low power, efficient. The Apple apps will be optimsed for the hardware as usual.

But for Audio users, we could see some very poor performance vs intels established architecture. Add third party DAWs like Cubase into that environment, and who knows what to expect, and even if we ‘matter’ enough for Apple anymore to be a prime considering for their hardware/os combos.

For example, There’s instruction sets such as AVX that a few plugins require now, i’m guessing ARM is going to support these off the bat? Or will that run through an emulation layer? That’s just one example outside of performance to consider.

On the flipside of all this, Apple could be taking computing down a far more streamlined and efficient route which will yield far greater results than us relying on Intel to keep squeezing and bolting more and more out of their existing architectures. And as a result, Audio users will benefit greatly.

It is depending on if it compiler generated instructions or handwritten assembly code. But it is a lot harder for a Risc CPU to do it than a compiler generated optimization.

The whole design of ARM means that you don’t have a ton of specific instruction sets, so I really don’t understand this question. Of course you won’t have any implementation of AVX there in any shape or form. The amount of performance estimation about an eventual ARM Cubase you can draw from this is exactly zero.

If you’re referring to the fact that Rosetta 2 doesn’t support AVX, well, I sure hope no one is going to run any DAWs through that!