Apple Releases M1-Powered Apple Silicon Macs, macOS Big Sur Releasing This Week

[birdie]

Uhhh....you do know there is Qualcomm, Mediatek and Marvel right? They are going to look at this M1 and use it as their benchmark in engineering and performance going forward.

Make no mistake. Qualcomm right now is looking to poach an Apple Silicon engineer as Intel did with AMD in order to make GPUs.

Apple has set the benchmark and the bar VERY HIGH in terms of ARM engineering and performance. There are LOTS of jealous engineers right now looking to do better.

Some of those will make WAY more performant ARM CPU's for the likes of Google and Microsoft along with Lenovo, HP and Dell very soon now.

I want to believe it's going to happen but the leaked results pen the yet to be released Qualcomm Snapdragon 875 at ~1150 GB5 points while the A14 shows massive 1600 (i.e. 40% faster). The rift is just too wide to be closed any time soon. Apple engineers have managed to do something with the ARM architecture which neither ARM, nor Samsung, nor Huawei are able to replicate. I don't even want to talk about MediaTek which has stopped developing its own cores a long time ago and now they use off the shelf ARM designs.

[Jumbotron]

I was wondering about this after seeing your comment, if Apple had a small line about their new PCIe 16x graphics cards to compete with Radeons or Geforce. But Apple seems to be going the other direction with M1.

Graphics cards are not allowed - the support isn't built in:
https://techcrunch.com/2020/11/10/ma...support-egpus/

I suspect this means the M1 will only be in the low and midrange AIO macs, minis, and laptops. I find it a little hard to believe Apple is going to stop selling its ten thousand dollar Mac Pro with twin Radeon Vegas in it.

Hopefully someone else can add in more information than I am aware of, or correct an error if I made any.

You keep forgetting along with most folks. Heavy metal CPUs and GPUs are worthless without software. That software has to be TIGHTLY tuned to the hardware to get the most bang out of it.

Every x86 based Operating system....every...single...one....Windows, Linux, Unix....has to be written for the lowest common denominator. You quite literally have a combined SEVERAL thousand SKUs of combined Intel CPUs, AMD CPUs, AMD GPUs and now Intel GPUs....and of course Nvidia GPUs. EVERY last CPU, GPU SKU has different capabilities and hardware peculiarities that requires software engineers to write SPECIFICALLY for each one at times. The OS's have to pack MORE and MORE and EVER MORE code into the kernel to make sure they are covering all the bases of these thousands of different SKUs. Which means bloat, instability and lower and lower performance. Which means you have to have bigger and bigger metal (power hungry powerful CPUs) to make up for it.

Not so much Apple.

They are COMPLETELY in charge of all aspects of the IP and Silicon chain. The number of Apple Silicon SKUs....single digits and probably will still be even after they introduce the first Mac Pro next year with ARM based Apple Silicon SoCs.

Their software and OS kernel are COMPLETELY tuned for their silicon in a way that Microsoft and Linux and Unix could and will NEVER be.

Those efficiencies ALONE will allow Apple to extract every bit of compute power from chips that at least on paper look less powerful than heavy metal x86 chips.

Also...remember...Apple Silicon now at least with the A14 based M1 is zero copy HSA. That data transfer architecture that AMD championed and actually succeeded in the mid 2000's and has now abandoned for now, is an integral part of getting more performance out of their silicon as opposed to others' APUs.

It's quite dismaying, actually to see so many people on this site who claim to know so much about how Apple sucks, and their silicon suck, not realize that compute performance is a holistic engineering feat that incorporates not only CPU raw power but interconnect tech and data transfer performance to also being able to heterogeneously and efficiently transfer compute loads across the die to the proper accelerator.

x86 Big Metal guys are the equivalent of Drag Car racers who tout their straight line speed. But the world has moved on to needing an F1 car.

[ezst036]

Uhhh....you do know there is Qualcomm, Mediatek and Marvel right? They are going to look at this M1 and use it as their benchmark in engineering and performance going forward.

Make no mistake. Qualcomm right now is looking to poach an Apple Silicon engineer as Intel did with AMD in order to make GPUs.

Qualcomm, Mediatek or Marvel know who they need to try to hire if they want to get it done.

Jim Keller.

[Jumbotron]

I want to believe it's going to happen but the leaked results pen the yet to be released Qualcomm Snapdragon 875 at ~1150 GB5 points while the A14 shows massive 1600 (i.e. 40% faster). The rift is just too wide to be closed any time soon. Apple engineers have managed to do something with the ARM architecture which neither ARM, nor Samsung, nor Huawei are able to replicate. I don't even want to talk about MediaTek which has stopped developing its own cores a long time ago and now they use off the shelf ARM designs.

True to a point with Mediatek. But have you read anything on ARM's "Neoverse" Initiative. And now their N1 and V2 archs? ARM basically is now playing catch up to the custom silicon work done by the likes of Qualcomm and Apple.

Some of this has been driven by Fujitsu's work on their world fastest Supercomputer. Some of this has been because more and more National and Continent wide Supercomputing and HPC initiatives such as Europe's Continent wide Super initiative requires a more "generic" CPU design that a custom built one like Qualcomm and Apple.

ARM going forward is competing heads on with Qualcomm and Apple Silicon. Folks like Mediatek that will just go the cheaper route and let ARM do the heavy lifting will still benefit from ARMs higher level of competition.

ARM is not only competing against the x86 cartel of Intel and AMD. They are competing against their own clients. This is FANTASTIC for the whole ARM ecosystem. What we are about to see going forward from the entire ARM world is going to be astonishing.

[Jumbotron]

Qualcomm, Mediatek or Marvel know who they need to try to hire if they want to get it done.

Jim Keller.

Very true !!

[polarathene]

Shall...I...continue?

Can Apple use OpenCL? OpenGL? (excluding unofficial third-party solutions) or CUDA?

Can I upgrade any of the hardware I'd use within a regular PC, like RAM, storage, GPU?

When the M2 comes out, can I upgrade my mac by swapping out the processor like AMD supports for multiple generations without having to upgrade the motherboard?

How long will my OS support upgrades for with Apple? When can I expect to be prevented from upgrading? Will upgrading remove features or industry standards commonly used?

Does Apple support developers from other platforms to build software for Apple products without requiring ownership of Apple? (eg can I build my app for windows and macOS on my Windows or Linux machine?

Can I run Apple OS via VM legally without running it on Apple hardware?

Has the situation improved where basic / default features on both Windows and Linux (eg extracting a single file from a zip archive) can be done without paying for software? Pretty much all the issues I ran into with macOS, the solution from Apple communities was to spend more money, or any software perf issues were fake news (terminal taking seconds to open before responding to input, UI updates prioritized over input responsiveness under I/O pressure).

Is support for running other OS on Apple hardware improving? Or is it declining? Basically, if I care about anything not endorsed by Apple on their products when it comes to running other software or common APIs, am I going to be disappointed vs on equivalent hardware running Windows or Linux?

What's maintenance costs like with Apple hardware? If/when something goes wrong, am I paying more or less money to get that sorted out and fixed? Is my options for support as good as it is with hardware running on Windows or Linux? What if I'm in a rural area with no Apple Store for many hours if that, but a variety of other tech stores around? Will it be a better experience?

What sort of choice and flexibility do I have with Apple products? If I want to add 128GB of RAM, how does that compare cost wise?

And the always fun question, for the equivalent amount of money, hardware wise, what value for money am I getting with Apple vs spending on a hardware for a Windows/Linux system? Best argument you have there is for non-technical / casual users that don't want to think about hardware and are happy to throw excess amounts of money if it means reduced chance of bad experiences.

and compatibility with more peripherals than ever.

Hilarious. How's that eGPU support? Bet I can't use CUDA still right?

How's the compute power of Apples hardware vs workstation grade NVIDIA GPUs that dominate at ML? The figures you shared are meh compared to dedicated hardware. If you can't expand compute power via eGPU (not just NVIDIA as AMD/Intel can be great for this too), big thumbs down. I'm assuming that you can and it's just NVIDIA that Apple is being a dick towards (probably won't making proprietary software vendors that exclusively use CUDA atm have much interest in accommodating for Apple though, so they lose those professional industries).

The World’s Best CPU Performance per Watt

The comparison is "Latest PC laptop chip", how vague.. The identify the highest CPU performance as 2x against this CPU at 10W, not exactly clear if this is single threaded or multi-threaded performance either, you can see how that can be played to their benefit by being ambiguous about such details right?

It is ARM based, so I am not surprised of it having better power efficiency over x86. I know of a 6-core that uses a similar power profile (but is definitely not 5nm, nor likely competitive in performance). The 8 cores themselves aren't equal, half are power efficient, while the other half are performance focused, would be interested in how that influenced results too.

The next slide / graphic looks at previous mac generation CPU single threaded perf, at cites 3x perf per watt improvement. Again vague what specific CPU that's being compared to, but not entirely surprising considering differences.

The powerful 8-core CPU performs up to 3.5x faster than the previous generation.

I love how vague this is. You're the Apple fanboy, so by all means correct me if I'm wrong, quick look shows the 2019 macbook airs with two different models at least offering either:

- 1.6GHz dual-core eighth-generation Intel Core i5 processor
- 1.1GHz dual-core 10th-generation Intel Core i3 processor

Probably safe to say the comparison is against the weakest option? Either or it's comparing 2-core to 8-core, so is a 3.5x faster metric here all that impressive really?

[polarathene]

For those who want to get a pristine MacOS X experience and ultimate performance (Intel Tiger Lake and AMD Zen 3 CPUs are nowhere near close in performance as M1 runs emulated x86 code faster than Intel/AMD CPUs run it natively, and RDNA2.0/Ampere are just laughable)

Apparently this post has been edited since? Was going to ask about sources on M1 running emulated x86 code faster than Intel/AMD natively, or if that was just comparing to performance of prior macbook Intel CPU, in which case, that's not a very accurate comparison/judgement. And to compare to RDNA 2/Ampere GPUs as laughable? Sorry but Apples M1 iGPU is not in the same league as those dedicated GPU compute power...

The A14X chip is already as fast as Zen 3 while being up to five times more power efficient. Consider its power envelope of less than 3W per core vs. Zen 3 cores which consume up to 15W.

Where is the 3W per core stat? (not debating it just curious as a quick google didn't seem to bring up results about it)

There was rumour of A14X being almost competitive with i9-9880H, comparing that to 8-core Zen3 5800X would suggest it's not as fast as Zen3?

https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9880H-vs-AMD-Ryzen-7-5800X/m750169vs4085

I did come across this which says it's 8-cores with only 8 threads, and 1.8GHz base clock and 3.1GHz turbo (yes I know that's not quite enough to go by, but you don't really get IPC metrics in specs):

https://www.cpu-monkey.com/en/cpu-apple_m1-1804

There are some results there that say it's quite competitive, but I'm a little skeptical (Only seems to have GeekBench scores). Would be good to see once more widely available. I won't buy Apple products personally, but welcome the hardware improvements and competition

[chuckula]

Can we ban Jumbotron? I mean, Phoronix is supposed to be a site dedicated to Linux hardware. He has spent every single post I have ever read talking neither about Linux and almost never even about hardware that runs Linux. There are plenty of Apple circle-jerk forums he can go to to feel better about himself since he clearly has no technical understanding of what Linux is.

[ezst036]

You keep forgetting along with most folks.

I haven't been having a performance argument along with the rest (or at least many) of the participants here. I haven't forgotten. The key is, I don't care.

Logically, the profound nature of optimization can carry hardware a long way. It makes sense to see how that could make up many differences.

They are COMPLETELY in charge......

Yes, they are. But that isn't a feature. That's a bug.

It wouldn't matter if the M1 with its optimizations could turn a threadripper into a trashripper. I detest Apple's control. They don't just control the optimizations, they don't just control the hardware design, and they don't just control the operating system and other software. Apple seeks to control you. They seek to control me. And it's disgusting. I'll gladly give up a little perceived performance to keep from having to deal with "how do I get around this" and the built in planned obsolescence.

I upgrade when I tell me to. I don't upgrade because Apple told me to. Apple is not COMPLETELY in charge after all.

[Jumbotron]

The take away from Anantech's review of Apple Silicon's M1 ( which as they stated is an Apple A14x )


Performance Leadership?


Apple claims the M1 to be the fastest CPU in the world. Given our data on the A14, beating all of Intel’s designs, and just falling short of AMD’s newest 5950X Zen3 – a higher clocked Firestorm above 3GHz, the 50% larger L2 cache, and an unleashed TDP, we can certainly believe Apple and the M1 to be able to achieve that claim.

This moment has been brewing for years now, and the new Apple Silicon is both shocking, but also very much expected. In the coming weeks we’ll be trying to get our hands on the new hardware and verify Apple’s claims.

Intel has stagnated itself out of the market, and has lost a major customer today. AMD has shown lots of progress lately, however it’ll be incredibly hard to catch up to Apple’s power efficiency. If Apple’s performance trajectory continues at this pace, the x86 performance crown might never be regained.


Here is the entire article from Anandtech. It is deep and it is shocking. Must read !

https://www.anandtech.com/print/1622...-a14-deep-dive