Apple M1 Pro and Max Surprises
mondaynote.com
mondaynote.com
Apple surprised people by what these actually turned out to be.
They’re better than folk expected.
We’ll have to wait and see how that plays out in the long run, but I don’t really think it’s something to get worked up over.
Yay, new cpus that don’t suck on a laptop that doesn’t suck.
I know it’s a low bar to clear, but it still makes me happy.
The m1 stuff has "tensor" cores and allegedly a faster GPU - that is, it's faster at some things my GPU handles for me, such as h264/h265 encoding.
Why should I get rid of my laptop and use the apple instead?
Ignore the 140W PSU!
Another question is whether the undoubtedly much better screen is worth the >2x price tag. You can get a nice portable monitor and a very good stationary monitor for just the price difference.
For me, yes! It's how interact with the machine for hours and hours each day. This is also why the keyboards over the last few years were such a travesty - they are the primary input method.
1. https://www.apple.com/shop/buy-mac/macbook-pro/14-inch-space...
configured with 32GiB RAM and 1TB SSD for $2599
Wrong question. You're comparing different values. Even 5-10% productivity gain can offset the 2.6x difference.
It's really a very specific niche that's targeted here: namely mobile(!) video editing and -production.
There are very few use cases these days that benefit from a high-performance mobile (cannot stress this enough) solution.
So unless your specific use case benefits at all from local processing power, I'd argue that it'd be hard to even get a 5% productivity increase that cannot be reached by other means (better/bigger display, more ergonomic peripherals, optimised workflow, etc.).
I'm not denying the fact that even a 5% improvement may justify the price difference. I'm just questioning whether the improvement in performance can meaningfully contribute to that in most use cases.
I think this heavily depends on what you're gaining. I think for a lot of tasks or scenarios it's not worth it. Especially around things like video encoding.
For example if you're a solo worker and you need to batch render 10 hours of video and your computer is essentially unusable for that time you would do this action overnight. If rendering everything finishes in 5 hours instead of 6 hours that doesn't matter because you're sleeping. If you need to render a quick 1 off video that takes 10 minutes there's a bunch of semi-productive things you can do while that finishes and it finishing in 9 minutes instead won't make a noticeable difference.
Now you might say if you're a professional video renderer you'll want to render videos during business hours as fast as possible in which case if you're operating at that scale you wouldn't use your daily laptop to do the rendering. You would remotely render it on a dedicated server (this is how most "pro" places work). This way you can comfortably edit video #2 while video #1 is rendering.
Even 6 year old hardware that costs $800 all-in can give you sub-100ms code reload times in most Rails, Django, Phoenix, Laravel and Node web apps even with Webpack or esbuild running -- even running within Docker on Windows (WSL 2) or native Linux. Also all major code editors can run well without input latency. That's a decently large amount of "pro" web development use cases. Unless you're doing hardcore continuous compiling in a spot where you expect to do this sequentially on your main laptop I can't really see a 10% gain being worth a 2.5x+ cost increase for most dev work.
For example if you can edit at $your_target_resolution and everything is liquid smooth without dropping frames even with a decent amount of effects then any improvement is overkill.
I'm sure for crazy high end set ups (8k video, etc.) an M1 Max would be nice but for other folks editing 1080p or 1440p videos at 30-60 fps you can get liquid smooth editing even with ~5-6 year old hardware that costs $1,000 all-in (a bit more expensive now due to GPU prices being temporarily crazy in price).
The improvement only materialises in a very niche use case (e.g. primarily mobile workplaces with virtually no additional weight/space for things like mice).
https://www.cpubenchmark.net/compare/AMD-Ryzen-7-4800H-vs-Ap...
The rest is not strictly about M1/Apple, but I found technically superior hardware can be quite sluggish depending on your software stack.
You probably shouldn't. Your laptop is more than adequate for your tasks. It may even be better in some respects (you can play some AAA games if you want).
Another thing to consider is that it may be marginally easier on you to develop on an ARM machine if you're planning on deploying to ARM servers, though this depends heavily on your workload.
I bought a Zephyrus G14 a year ago with high stats, but I often get weird issues where I have to restart my computer or find some fix online. And there are some build/design quirks that are a little annoying. I already want to replace it.
I also have a Macbook Pro from 2015 that I only switched because I wanted a better GPU. It still works fine and looks great. I don't think the G14 will be so great after 6 years.
We also have Macbook 13" 2018. The famous butterfly keyboard on it got broken after a year. It was repaired cost-free, but it took for Apple reseller 2 weeks to do that. And then overheating was terrible even after a dust cleanup. It was necessary to use a cooling pad under a heavy use. When I got a newer laptop on my job I tried to sell it for a reasonable price, but it turned out the resale value was much lower than I expected. So we keep it as a sort-of spare/guest machine.
I agree. If you know how to hack, of course hardware lasts longer. I think i have a Motorola PPC[0] mac somewhere that boots BSD
[0] totally 603e
This is really the first time in years a purpose-built professional laptop has came out with nice build quality, keyboard, display etc that clearly prioritizes pro users both in form (it's fatter for cooling again!) and function (beastly ARM SOC).
So while it's a low bar it's a loooong time since it's been cleared.
Apple just had an extremely fallow decade in the 2010s (in terms of design) because they were making so much $$$ with iPhone that they stopped caring about the Mac. Jony Ive also left the company during that period.
I so hope there will be a Mini with the M1 Max, as this would finally be the desktop Mac, so many people have been waiting for (me including).
The future finally happened in terms of fast external peripherals (especially storage), and I don’t have a burning need for a Mac with upgradable internal storage anymore.
In 2010 I got a $100 graphics card that let my PC do things that the PS3 and Xbox could only dream of. Now for the cost of a “mid-range” GPU it’s possible to get a Switch with a game or two that look fine and run at fast FPS.
One thing: Macs need to cut the Bs and give us upgradable RAM back
I don't see that happen any time soon, not with their new focus on building the core as a SoC. Maybe with the MacPros, there will be on-SoC RAM and slower off-SoC, but I doubt it till I see it.
RAM slots aren't free; all that extra interconnect length increases power requirements and becomes impractical with large buses. You'd need 8 SO-DIMM slots to match the bus width of the M1 Max models, at much higher power consumption and slower performance. LPDDR RAM doesn't even exist in removable form for these reasons.
Modern embedded SoC integration is insane. Yes, the "two" memory chips you see on the M1 are really 8 discrete dies/chips (4 stacked together inside each of the two packages), and the "two" big fat things around the Pro are actually 2x8; double them up for the Max and that's how you get to 32 channels. Good luck putting that through a connector...
never going go happen as Apples loathes users' freedom. they prefer selling a disposable brick.
Yes. I remember seeing several Clevo models in their offerings. Like most laptops, they have really bad thermal management and atrocious manufacturer software.
We basically have a 5+ year gap of little improvement in PC Industry. ( Comparatively speaking ).
7 year gap, during which Apple also experimented with the trashcan, the emoji touchbar, the shallow keyboard, the non-magsafe non-SD connectors. It was like a pause on personal computing.
7-year gap from 2013 to 2020 where social media became political machines, and all companies became cloud-first.
Strange decade indeed. Unless you are in US and the financial disaster hit hard, then I think I look back in 2010 that 2000s were comparatively uneventful.
I seem to recall the future of Apple hung in the balance, and they were looking to acquire either NeXT (and Steve) as the foundation for the modern Mac - or Gassee and Be. Sort of a fun parallel universe of what could have been.
I honestly think Objective-C is a big part of Apple's OS success.
Edit:
Building system APIs based on message passing provides a lot of flexibility and capability without having to worry about ABI compatibility or object layout. Message passing is good for hiding implementation details.
Obj-C also provides what Swift now calls protocol-oriented programming. Define a protocol and any object no matter its type can implement that protocol and be, say, a delegate for a long running operation. Etc.
This allowed Apple to innovate in their OS, providing increasingly more functionality more rapidly than what seemed to be happening elsewhere.
I think there's even more to this, but I'm tired.
FWIW I've written frameworks in both Obj-C and C++ (macOS kernel)
> "Obj-C is a much better framework language than C++. How and why is that true?"
Anyone could throw these sort of claims about absolutely anything and doesn't elevate the debate at all.
> "I honestly think Objective-C is a big part of Apple's OS success."
How and why is that true? Same thing.
Once upon a time, there was a company called Taligent, which was created by IBM and Apple to develop a set of tools and libraries like Cocoa. About the time Taligent reached the peak of its mindshare, I met one of its engineers at a trade show. I asked him to create a simple application for me: A window would appear with a button, and when the button was clicked, the words “Hello, World!” would appear in a text field. The engineer created a project and started subclassing madly: subclassing the window and the button and the event handler. Then he started generating code: dozens of lines to get the button and the text field onto the window. After 45 minutes, I had to leave. The app still did not work. That day, I knew that the company was doomed. A couple of years later, Taligent quietly closed its doors forever.
It is asserted a bit later than C++ was the problem.
I get the feeling some of the Java Cocoa experience applies to C++. Although the C++ of the 90's is most definitely not the C++ of the 2020's.
I often wonder what if Apple continue to improve Obj-C instead of Swift.
I really like Alan Kays's metaphor. Instead of reinventing the wheel ( which isn't so bad ), more often than not in technology they reinvent the flat tire.
For example, CoreFoundation has a C API... that would be a bit of a test of a C-only framework world.
You can see however that's missing the convenience of Obj-C.
I hope Swift will be successful. It is nice to use (most of the time).
If you think that for a long period of time all apps were built in Objc for Mac & ios, and how quickly they managed to make an ecosystem of developers who were building applications for Apple systems - I'm pretty sure if they had a c++ framework that wouldn't have happened so easily.
As an illustration, there's a video of SJ building a quick GUI app to demonstrate RemoteObjects (in response to MS's Distributed OLE) where he quickly knocks up a slider and text box. There's also another video where a NeXT developer goes head to head with a Sun dev in a race to finish a simple gui app (you'll never guess who wins).
Why did SJ and NeXT go all in for RAD tooling and making it easy to build apps on their platform? Because of feedback from his days at Apple when developers complained it was too hard and slow using the MacOs toolbox APIs.
FWIW I also believe that a large part of Microsoft's success was due to Visual Basic, also a hugely successful RAD tool that brought thousands of new developers to its platform.
It's like a government during an oil boom. They can get away with a lot of shady stuff as everyone is feeling the wealth. It's only much later that you can soberly analyse how much was stolen from the people and figure out if you are in Iraq, Venezuela or Norway in terms of the benefits of Moore's Law and how that got spread around the economy that made it possible. If there's Billionaires anywhere in the picture then something probably went wrong.
I suppose it's natural to cheer on the technology leader - and Intel is and has been a serious company who has invested a lot to move fab technology forward.
The efforts to build a consumer brand started by Andy Grove have probably also helped quite a bit.
Nevertheless, its striking how little focus there has been on Intel's market share and the lack of real competition on the desktop and server. Good that that's changing.
https://www.extremetech.com/computing/184323-intel-stuck-wit...
It didn't have to be this way. We could have had a good AMD ARM platform instead. But they, like Intel, had become complacent (and this despite their position as fighting for Intel's table scraps in the x86-64 world). The failure to take risks leaves us all worse off.
Intel needs to exist, if only to prevent a TSMC monopoly at this point, but what good is AMD if they can't even compete with Apple when outsourcing to TSMC?
With that pressure they cannot sit around and take a risk on a long term unknown that did not guarantee returns. Maybe today they can pursue this path(I bet they are experimenting internally). I recall some quote by Lisa Su that they are open to other arch long term.
I know it's shocking that the company that produced in-house ARM chips that are top performers in the mobile space for the last 10+ years - has made a bigger chip and shockingly(!!!) it's good.
So, the biggest company in the world has created a semi-custom closed ARM design, running on their properiatery Operating Systems - iOS/iPAD OS/MacOS, that run only on their absolutely unable to repair/upgrade machines, that are essentially bricks of alliminium with a display.
Me personally have a
ASUS ROG STRIX G15 laptop with Ryzen 9 5900HX and 2 x 32GB DDR4 3200 of memory - I can run ot it Windows, Linux, MacOS, BSD, etc.
99% of the time it's connected to the power outlet and I can overclock it.
I don't get the Apple Sillicon excitement, tbh.
It will be fast enough on their machines, but they will forget some people don’t like/can’t afford these machines.
But the trend towards on-chip integration and SoCs has been decades long. CPUs also used to be made from discrete components. So I think, for better or worse, the transition to SoCs is inevitable.
This.
Now go outside and try this laptop as a mobile device.
What's the point of comparing what is basically a workstation to a relatively thin mobile device?
>I don't get the Apple Silicon excitement, tbh.
You get very power-efficient mobile device. If the mobile part is of no interest for you - than there is nothing to talk about really.
It would join the very short list of usable reverse engineered GPU drivers, and not for lack of attempts. Just look at noveau for an idea of what to expect at most...
Oh come on.
Several years ago, Nilay Patel of the Verge reviewed an Apple product (think it was a phone) and lamented that it didn’t have style. That comment unleashed a wave of memes and ridicule towards him because he was known for wearing a spike bracelet with every outfit.
Well, I found the hoopla hilarious and jumped in a back and forth Patel was having with the author of this article. I was promptly blocked and still am to this day. I forget but quickly remember when I follow a link via Slack.
Totally off-topic ramble, carry on.
Amazing.
That image came straight from apple's keynote.
Yes these machines are used by lots of video editors etc but it does seem to be symbolic of a general shift towards more compute being done on the GPU - now even on laptops - and not just for video.
And having more GPU compute available at your fingertips will only reinforce that trend.
There's a substantial subset of high-performance computing tasks that are limited by GPU performance. But that's far from the whole equation.
trying to work out if this malapropism was used ironically...
The key thing in consumer silicon chip design is always individual core design (or designs, there's Firestorm and Icestorm in big little). That's what matters for single threaded workloads.
Fully parallelized workloads in a consumer laptop are just exceedingly rare. Plus it's exceedingly pointless to do lengthy multi-threaded development on a laptop, it's going to overheat and under-clock. SSH into a cloud instance and develop in an identical environment to where you deploy.
Apple vastly exceeded expectations by making enormous silicon dice, and tying in vastly more GPU, bandwidth, and parallelism than expected. To the joy of video editors everywhere (which cannot be done over ssh).
But for people who watch closely, the M1 Max delivered exactly what was expected, plus a gargantuan GPU and memory bandwidth.
The M1 Max is still ultimately the A14's Firestorm/Icestorm cores, in a MBP form factor, delivering amazing idle and low power consumption performance, and single threaded that matches Intel/AMD's desktop chips, at a far lower PPW because of the node shrink.
Given that the A15 still features the Firestorm/Icestorm core design, it's almost guaranteed that we'll have to wait till 2023 to see a more than marginal uplift in single threaded workloads and low power performance.
Really? I've observed these kind of consumer workloads getting more and more common in recent years. In any case I wouldn't call them exceedingly rare.
Maybe I misunderstand how many people do streaming TV with downconvert.
But that said, very chuffed OB does ffmpeg. You rock!
It's very common to multi-thread something across three threads for example.
But something that is perfectly parallelized across all your threads, or at least 8 or more, that's actually something I almost never see in end user devices.
And yeah, ultimately having 4 "big" cores that are as fast as possible matters for those use cases that are threaded across a limited number of cores.
The A15 uses Avalanche performance cores and Blizzard efficiency cores. I'm not sure how different they are, except that Blizzard in particular is a lot faster that Icestorm.
The high performance cores are very similar iirc, accounting for binning and density. But if Apple has upgraded just the efficiency cores this time around that makes sense.
Eh? Probably THE most common workload for a laptop (opening a webpage) is multi-threaded. And even consumers multi-task. As soon as you have multiple processes, you can use multiple cores even without multi-threading support.
> Plus it's exceedingly pointless to do lengthy multi-threaded development on a laptop, it's going to overheat and under-clock.
The promise of these laptops is that they won't. As you say, their power consumption is a lot lower than competing chips, and it's that power that produces the heat.
There, I fixed it for you.
People who are used to Apple laptops and like using them, will love this new laptop. People who did not look favourably upon Macbooks, still have no reason to, except maybe battery life. But even there, AMD laptops are reaching 9-10 hours easily, which may be more than good enough for many users.
Anyway, I don’t think it’s surprising that the company which has one of the best global production and logistics operations in the world is capable of building good processors. It’s just another part of the puzzle after all.
It’s just a shame that their design department still doesn’t seem to have learned it’s lesson from Jobs in that it once again has failed to build a product that is designed in a way that people want. Why they ever deviated from the MagSafe, why they added the weird screen Function bar and now why they’ve added a bezel to a ducking laptop is really beyond me? Like, do they not use computers? But improving upon the supply chain by creating new powerful hardware? Absolutely within what I would expect from modern Apple.
They were the first to remove serial ports, floppy disks, internal modems, optical drives... And it turned out good even if you could say that's not what the people wanted.
The touch bar, magsafe removal, etc.. were failures alright, I bet the notch won't be a failure, however
https://arstechnica.com/gadgets/2021/10/samsung-tablet-leak-...
The notch is already being copied. Perhaps the wrong way.
The difference, and what I probably failed to explain well enough. Is that these removals might not have been what people thought they wanted, but what Apple knew they did.
With the touchbar, new keyboard, removal of the MagSafe, removal of the charger from iPhone boxes Apple has made things people didn’t actually want.
The notch on the phone may have seen wide adoption, but in my anecdotal experience nobody likes it.
But do you really think people would rather forgo the camera and other features to remove the notch?
Alternately, do you really think people would prefer a notch-high bezel at the top of their phones instead of more screen with a notch?
People are complaining about getting extra screen real estate.
I have one on mine and while I don’t particularly dislike it, I’d certainly rather not have it.
Which is sort of my point in Apple losing touch. They stopped changing things in ways that I ended up liking, and I use myself as a reference because I tend to be rather ordinary in my consumption.
Maybe my anecdotal experience is completely wrong, but I have never heard anyone who were happy about the notch on their phone.
If they made a triangle of extra screen area jut out on the right, you'd get extra screen space too. I bet people would still defend it not because "it's extra screen" but in reality because it came from Apple.
The fact that people are so willfully ignorant of that is amazing to me. The Apple-think bubble is stronger than I thought.
They could have just not extended the screen and had a larger bezel. I don't see the issue.