Seems like Arm opened a new world when it comes to computing with Mac, what do you think about the future of software and applications?
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.
Are Applications/Software going to Improve or Disimprove with new ARM Architecture?
- Thread starter trevpimp
- Start date
- Sort by reaction score
In my guess it opened up a new world by making it easier to access information both readily in code and software? Im new to computer logic/code so Im asking questions
On the basic "what can the software do" level, there is no difference between x86 and ARM. They are basically just two different takes on how to encode computer operations, but the end effect is the same. Modern ARM does have a bit more features that can benefit niche applications, but so does Intel. From developers perspective, nothing much changes when moving to ARM.
There are other, more signifiant factors however. First, by utilizing ARM in all their devices, Apple makes it easier to share code between the iPhone and the Mac for example. This makes testing quicker and cheaper, and reduces the chance of bug occurrence. Second (and something I find very important), Apple's devices are not just about ARM CPUs — they introduce a specific architectural model that Apple calls Apple Silicon. An Apple Silicon device is a collection of heterogenous processors (including an ARM CPU, Apple GPU, Apple ML accelerators etc.) that share the same memory. And such device offers unique features that indeed allow new software to be developed. For example, with an Apple Silicon Device one can write video editing software that combines the GPU and the machine learning coprocessor to apply effects to the same video stream practically simultaneously.
To sum it up, ARM processor on its own does not mean much. It's the overall Apple Silicon architecture that will allow some really interesting applications to emerge.
Wow, you make really good points with new information to my knowledge
Thank you for your kind words!
I noticed my x86 apps to be faster being translated by Rosetta actually, especially apps based on Eclipse like the HDB Studio.
Honestly I have 0 complaints about apps that I'm still using via Rosetta, right now it's only 3 of them. Teams, Citrix Workspace and VPN.
I'm trying to migrate all my x86 apps to Citrix anyway
Honestly I have 0 complaints about apps that I'm still using via Rosetta, right now it's only 3 of them. Teams, Citrix Workspace and VPN.
I'm trying to migrate all my x86 apps to Citrix anyway
Judging by the past almost 80 years of computer programming, more power also breeds complacency in developers: It allows the worst of them to get away with sloppily written code and ridiculous abstractions as their customers just purchase even faster computers to run their programs. This is a function of the human nature that won't disappear just because our computers get better.
Nice post. Only one thing - the CPU is Apple Silicon also. It is not an ARM reference design like Cortex.
The tools used to create the applications remain unchanged. The only difference is they will produce ARM-compatible binary code instead of x86-compatible binary code. I somehow fail to see how this should have any implications on application quality whatsoever.
Software will get better, because Apple can add SDKs that depend on specific CPU features that Intel wouldn’t or couldn’t provide. Think of all the things your iphone/ipad can easily do that your mac cannot, because of hardware differences. Over time, this will become more important.
Intel vs PowerPC is comparing Apples to Apples. M-series to Intel is Apples to Oranges. The PowerPC and Intel chips back then didn’t have any special purpose computing hardware on them. They were just CPUs. Now you have Intel selling CPUs and Apple making chips with heterogeneous computing cores on them. Apple has already added some circuits to M1 that have no equivalent on Intel chips, and that is going to continue to an even greater extent going forward. This will provide software developers with capabilities they won’t have on Intel, and it’s more than simply a speed bump or power reduction.
It's the hardware that will enable more powerful software to run where it previous couldn't. Already the MBA is running apps with performance that was unheard of when using Intel chips. There is potential for other vendors to create custom SoCs as well to enable new form factors, though they will be more dependent on getting OS support from MS or Google.
Hopefully we'll also be returning to an era where software was more portable and able to be used across different chips and architectures.
Hopefully we'll also be returning to an era where software was more portable and able to be used across different chips and architectures.
It's the same underlying instruction set (ISA), so other than the Apple-specific stuff built on top of the core ISA, the instruction set is the same that Qualcomm uses with their Snapdragon SOCs (including the Surface Pro X), MediaTek SOCs used in lower-end Chromebooks and Android handsets, and other manufacturers of ARM-based processors use for their processors/SoCs. Apple Silicon is more a generic term used to distinguish Apple's SOCs from x86 and the rest of the ARM market as well, but it is still at its core an ARM-based SOC.
When you have the M1 Macs competing head to head with both 15" and 16" MacBook Pros from 1-2 years ago (and even the Mac Pro in some scenarios) despite having less memory and being the entry-level devices in the Mac lineup, I think it's safe to say that Apple has already lapped the field in comparison to either the 64k-PPC or PPC-Intel transitions. That's on the performance side of things. It is still up to developers to rewrite/recompile their apps to take advantage of the new features and possibilities offered by the switch to the M-series of SOCs.
Agreed. Their compilers are optimized for the M-series now and so is their software. Perhaps even more so now that some metal APIs can start to wor better in my opinion.
The ISA is really far down the list in terms of importance - micro architecture is a much bigger factor in performance and efficiency And Apple Silicon’s micro architecture is unique to Apple. Indeed the fact that Apple Silicon uses RISC is even more important than the ARM ISA.
Funny thing is, if they were using CISC, but a CISC with fixed-length instructions, the benefit of Arm would be much reduced. Maybe even eliminated (if instruction fetch benefits of this hypothetical CISC outweigh decode penalty).
The other issues with CISC - funky addressing modes, paucity of registers, etc can be compensated for much more easily.
Any RISC ISA will be better than x86 largely because of the difficulty in decoding x86 caused by variable instruction length (which, in turn, causes difficulties in instruction scheduling and parallelism).
It provides an opportunity for more bloat to slow the fast ARM chip down until it becomes unbearable. Apple macOS gets worse with every release. Right now Big Sur is a heap of trash so goodness knows what it’ll be like in five years time.