The Apple A15 SoC Performance Review: Faster and More Efficient
anandtech.com
anandtech.com
Do people really expect revolutionary improvement every single year? ~20% Improvement YoY within the same power envelop is god damn impressive. And if this is not good enough? This is just layering ground work for next year's 3nm and LPDDR5. ( Or 4nm depending on circumstances )
The E Core are interesting because they are a preview of what to expect in next generation Apple Watch.
I was hoping there is some investigation on new Display Engine and Video Encoder / Decoder. Especially on power usage. But looks like these kind of interest are in the minority.
It’s just a knee jerk reaction to about anything related to Apple. There is a lot to dislike about them, to be sure, but these reactions to independent benchmarks are just blind cynicism. It’s cool to be a contrarian, and it is making the Internet a much worse place.
Imagine having a desktop setup and an "empty" laptop case with slide in dock.
Everything in one place. Same system for the iPad.
Will probably never happen though because that's less products to sell?
1. No one used Windows Phone either
2. The use-case wasn't really clear.
"Run Office apps on a regular computer-sized monitor!" I can already do that, using the computer-sized computer connected to my computer-sized monitor.
"Setup is simple using Continuum-compatible accessories" So I have to buy a new keyboard/mouse/monitor/something else? (No, but it sounds like I might have to)
The pitch was "You can use your phone kind of like a PC, assuming every app you need runs on Windows phone" (which, for almost everyone, is absolutely not the case). If there's one thing Windows phone was famous for, it was not having apps.
Essentially, Microsoft was trying to pitch you a feature that let you use a phone + monitor and keyboard as a replacement for some of the apps on your PC. Not a compelling pitch, except for presentations.
Circa 1998-1999 I was using a mobile device with a touchscreen and wireless broadband to surf the web and send email while sitting outside. Clearly the idea was good even if Handspring and Ricochet never managed to deliver it with few enough compromises to see mainstream success.
https://en.wikipedia.org/wiki/Samsung_DeX
makes sense when these devices are already more powerful (and expensive) than you Raspberry Pi.
To think they are just casually outperforming their desktop chip in the benchmarks. Man I can't wait for the next M1X/M2 or whatever they call it.
No, thank you, I'll keep a flawed x86 architecture. We could also stop writing shit software.
The main difference that I see is that the M1 chip is much faster and more power efficient than the systems it replaced (as well as much more expensive systems, for that matter; see M1 Air vs. Macbook Pro). I suspect that means that people who buy the M1 systems that Apple is selling won't replace their laptops as soon as people who bought x86 versions.
Your point is taken, but this is a trend that Apple has been moving towards regardless; it's not something that's new to the M1 Macs. You haven't ever been able to swap out your motherboard or processor in a Macintosh with very few exceptions (including the Xeon Mac Pros), and we've yet to see what an M1 (or M2 or whatever) Mac Pro would look like, expandability-wise, anyway.
I used to think like that - until I analyzed all my PC builds over the last 12 years and realized with Intel changing sockets every generation the whole incremental upgrade thing was fond memories from the 80's, 90's and early 2000's. Once I realized incremental upgrades wasn't a hill worth dying on my choices increased and I've never been happier with the size/weight/speed/battery life that previous dogma prevented me from even evaluating.
>We could also stop writing shit software.
We could but people don't. Sentiments often don't survive contact with reality :p
To this day there still exists a cadre who think Apple users are idiots for spending so much money on a computer. Many of them have not seriously taken a look at what the platform offers and they do not fully comprehend why people choose this platform over others.
Regarding the "We could also stop writing shit software." quote, it seems as if there has been a wave of optimization over the last decade just out of necessity. (electron notwithstanding)
It is not enough, components that are written in fast languages such as the OS have generally gotten slower. We are forced to upgrade due to security and I don't see any incentive for companies to improve this part of the software stack. You can use other desktop apps that are faster but you only really have 2-3 choices in OS and that dictates everything else you can do with your computer.
Was running Windows 11 on the Parallels Desktop on M1 and it seems snappy but not as snappy as MacOS. I noticed this same behavior on Windows 10 running on a core i7. It seems as if the design of the OS has some bloat that I cannot explain clearly but can definitely feel. We should consider that maybe Apple has been also optimizing software to be more performant.
Alder Lake 2022.
They absolutely do. You can see it in all their bundled programs, too (except Xcode... sigh). Safari, Pages/Numbers/Keynote, Preview, Maps. All snappy and light on resources. None of them prone to slowing down unrelated things on your system. You can leave them open in the background for weeks and forget they're there.
There's also iOS and Android. Android's gotten better, but it's always needed more energy and faster hardware to approach iOS performance and UI responsiveness. The software's simply better, by that metric, anyway.
Another instance of this was visible in BeOS vs. Windows/Linux, back in the day. The difference was not subtle and was present even when running cross-platform programs, so it must have been BeOS itself (and its libraries, SDK, et c.) responsible for everything feeling way faster and more responsive.
A significant portion of this is just requiring more efficient developer tools.
- Generational GC vs ARC memory management accounts for a large (50-100%) increase in memory usage
- JIT vs compiled code performance impacts
- Swift value types vs Java (and Objc) class types removes further memory management pressure as well as allowing for significant memory savings.
- The underlying OS model Apple has restricts the need to run code from multiple applications at one time to a small set of extension points. Android does allow for background processing (although you can't guarantee consistent/persistent processes).
Of course you give all this up when you are just wrapping your javascript + web content up in an electron or similar framework. In that case, you are running at a fraction of the efficiency of either platform.
Hardwired storage is horrible from a data recovery standpoint also, you should be able to take our your storage chip and try to recover data from it in a standalone fashion as long as you know the encryption key, even if your main board or CPU has been damaged.
Their storage and RAM prices being double the market price is also irritating.
Yep. I still have desktop machines and build another every 4 years or so. I used to upgrade machines, but it's just not realistic anymore. If I'm very lucky I'll manage to reuse the case and power supply, but even that's iffy. Stuff's changing too fast, including, as you mentioned, processor sockets. Memory, too. Man, I had some memory in the 2000s that lived through like 3 different machines. Not these days. If I upgraded video cards more often I guess I might manage to do that once per machine, but that'd be it.
Buying entire new computers is the most wasteful thing you could do, both money wise and environmentally wise.
There is definitely value in SOC for CPU/MoBo (maybe memory), but the other components work better upgrading at different rates.
They've been on the same couple sockets for a long time now.
Processors/Motherboards not as much, maybe once every 5 years or so.
Apple really seems all about throw away culture.
Then go for it - you have choices, and no one is forcing you to buy a Mac.
I personally got shit to do and have no interest in replacing all of that at that type of frequency.
Just because Apple computers before last a long time doesn't mean it will be the same with their current computers.
Although my personal bet is that the 2016-2019 MBP will degrade faster than the 2020 M1 Macs.
I have tried giving older Macs to other family members, that didn't work out so well. So there are limits.
In general, used Macs have crazy high resale value, and if they work out ok, no problem with that. Both ends of the transaction receive value from the exchange.
I am in the process of clearing out quite a bit of e-waste, but we are talking fried PowerPC Macs and some iPhone 4 devices that I cant get working again. I have noticed that the Macs tend to be physically smaller than thier counterparts from the same era in my collection; that might suggest less e-waste overall, but of course the variation is all over the place.
The oldest operational x86 machines in my setup are two tower Mac Pros, a 2007 one and anither from 2009. These are somewhat upgradeable: you can use SSDs instead of rotating rust (duh), max out the RAM (easy), put in newer GPUs (drivers and BIOS permitting), or max out the CPUs (you need a moderately weird tool to get at the heat sink screws).
The fate of our electronics is indeed worthy of concern, but I don't see that Apple approach leading necessarily to more trash.
(oh yes: I am typing this on an ipad 4. I'd love to keep using it for another 14 months, it will then be ten years old. but it's getting a bit unreasonable. Works for hacker news though)
My assumption your approach is far more wasteful at a cost of getting top notch performance. Nothing about the bleeding edge is ecological.
Just to clarify though I don't replace all those components every year, it usually ends up being one or maybe two things a year. But if I do one component a year my system can end up lasting indefinitely.
If you can get by on a decade old machine all the more power to you. Professionally I cannot.
It still received the most recent OS updates.
Battery life is not as good at 8yrs - but still has a few hours of juice in it.
Overlooking the butterfly keyboard fiasco, Apple makes robust long lived hardware, and software to support it.
It’s market does not include the 3% “piecemeal” upgraders. But I suspect in “amount of time component is use” metrics it handily beats the wintel world (and the Android world)
A15 is also on a newer process node (TSMC 5nm gen2) and has 2x the L2 cache it looks like.
Given Apple's history, I wonder if like 10 years from now they just drop this additional bloat. In that case we really didn't lose the performance just gave it up temporarily. :)
The Windows 10 ARM beta also did a shockingly good job of running X86 code in Parallels too. Way faster than one would assume is reasonable for what was going on under the hood!
When you think about it, the M1 perfectly encapsulates Apple under Tim Cook. It didn't come out of nowhere. They have slowly improved on their A-series in the iPhones the last decade and it accumulated into the M1. That's exactly how Apple is run today. Inch by inch.
M1 powers by far the best arm based laptop available today - all arm laptops hence forth will be compared to it. revolutionary.
Apple iterations on arm have been regular and consistent for a long time.
But it's not true. ARM based Chromebooks have been on the market for years already and cornered educational markets and had massive consumer success overseas.
Nothing changes in the way people use their computers. It's a performance leap, but there have been many such in the past. Back when PCs and laptops got 2 cores instead of 1, "real" multitasking was finally possible and performance doubled almost overnight. That was IMHO a bigger leap than M1, but I don't hear people calling Core 2 Duo and Athlon X2 "revolutionary".
It also delivers incredible battery life, and runs a lot cooler so you don't have annoying fans or face throttling as much, if at all, under load. No more cases hot enough to burn skin.
If you think you understand the M1 just by looking at benchmarks or reviews, you are sadly mistaken. Use one of these machines for a couple of days and you will more than likely not want to give it up.
Heck there are developers ditching their iMac Pro's in favor of M1 Mac Mini's for everything but long compiles. When Apple releases desktop versions of their SOC things should get stupid spicy. And I'm very much looking forward to it.
Then you might want to look up the definition. A revolution is "radical" and makes "fundamental changes in the socioeconomic situation".
Life after the printing press / steam engine / transistor fundamentally changed. Even if tomorrow, all x86 chips were replaced with ARM chips, what radical and fundamental changes to society would take place?
I submit if you need to qualify your use of the word "laptop" with fully eight adjectives or adjectival phrases to avoid ambiguity with pre-existing products, that it's probably not "revolutionary" by definition.
The M1 is a good CPU. It's incrementally better than the Intel CPUs used in earlier versions of the same product.
It is very notable in that it was produced in-house and not purchased from Intel. And that says important things about the business climate in which Apple finds itself. But that's not the same thing as "revolutionary" in a technological sense. It does what competing chips do, somewhat better.
I suppose everything can be called incrementally better by some viewpoint but calling it incrementally better than the Intel CPUs used in earlier versions is wild to me. I specifically switched to my first ever Apple product because of how revolutionary the generational step was compared to what we've been getting on x86 laptops year to year and no equivalent x86 laptops of the year came close. Not only is it comparable to my overclocked desktop in tasks from 1-4 cores it does so without kicking on fans and the battery life is astounding while it does it. It even runs emulated x86 software faster than the native x86 version of the same year.
It's not until you get to the highest end desktop CPUs from Intel or AMD that you see competition with the M1 on a per core basis.
Keeping in mind this is Apple's lowest end, low power, mobile part. It wipes out all but the highest end desktop processors - not just on a core basis, but even a multicore basis - agin for all but the highest end.
Nope, I don't think revolutionary is hyperbole at all. It also makes me wring my hands with glee thinking about what a part designed for the desktop where higher power and cooling are readily available. Where adding more cores makes a lot of sense. How big of a gap is there going to be with that SOC?
The prospects of new hardware are once again exciting. CPUs have been stagnant for a LONG time now. I was simultaneously relieved and disappointed to learn my i7-7700K was still not that far off from the latest CPUs - hardware unboxed had a great video https://www.youtube.com/watch?v=wAX1lh985do
At best - on average a 30FPS between my lowly 7700K and the highest end parts? Yeah, newer parts have more cores - but most games are not well optimized for multiple cores. On one had I'm relieved - no need to spend money on upgrades since it just isn't worth it. On the other hand - it isn't worth it to upgrade! And what nerd doesn't like having the latest tech?
So yeah, it's nice to have someone pushing the boundaries and something to be truly excited about. Also stark contrasts like this tends to light a fire under others, if nothing else. Rising tides raising all boats and all that...
This isn't even remotely true. Computers are really good at doing more than one thing at a time. I find it mind-boggling that there is such a focus on the performance of a single core. This hasn't been all that relevant for the last 15 years. As for what this matters, the difference for single core isn't all that big regardless (one M1 core performs about the same as one 5980HS thread). When you want to know how powerful the computing power of a CPU is, how about comparing .. you know... what it can do?
And, what it can do when it is asked to do everything it can, I can reiterate that the M1 is nowhere near the competition. The 5980HS mobile CPU will get the same multi-threaded job done in half the time, which is a huge difference. Which is why discussion threads like these feel... chilling. Like as if I stepped in to a cult meeting, where we praise normalcy as revolutionary. It's not even close to the last generation of CPUs in terms of compute power. It just a step up from the Intel CPUs apple used in the previous MBPs.
And, it's not that the M1 is a bad CPU. It's actually pretty good for what it does, and it is exceptionally good at power efficiency. The 5nm and low thread count does contribute significantly to the power performance, so it will be interesting to to see how future M1 processors perform.
Imagine if the M2 doubled core count, and thus doubled multi-core performance compared to the M1. It would then be just as powerful as the current 5980HS. Another revolution no doubt.
I think that other manufacturers will look into their energy usage as well. That's quite revolutionary, because it seems to me that it has shown a proof of concept for which there's a lot of room to grow in. Less energy means less heat, less heat means you can crank up clock cycles.
My thinking is simplistic but I think someone with more understanding of it will tend to agree with my high level view and tell you exactly why this is revolutionary.
Note: I'm not giving Apple full credit here, I'm giving it half credit. The other half goes to the invention of ARM and RISC micro-architecture in general.
This was the whole marketing push behind projects like Intel Centrino - to drive even lower system power consumption by mandating certain combinations of certain parts that worked well together. Which continued with things like Intel Atheno & Evo.
Nothing Apple did with the M1 changed the game. It's the same game, it's the same race it's always been, they just are now in the lead of that race with their car. Which is impressive in its own right, but definitely not "revolutionary"
> Less energy means less heat, less heat means you can crank up clock cycles.
This goes the other way around. To improve efficiency you reduce the clock cycles and increase IPC instead. That was the M1's advancement over the status quo, a significant increase in IPC. Clock speeds didn't change - in fact, it regressed by a tremendous amount. This regression in clock speed is how the M1 consumes so little power by comparison.
The first lightbulb isn't revolutionary, it's a nifty toy invention that doesn't affect anything, much less cause a revolution like the first practical efficient general purpose mass produced lightbulbs did.
The first radio isn't revolutionary, the first practical efficient general purpose mass produced radios make the revolution.
The same for computers, the same for smartphones - iPhone was revolutionary because it impacted the world; while its competing predecessors that had almost the same tech did not and so were not revolutionary.
I count five: It's the first "widely popular", "fully-functional", "powerful", "general purpose" ARM-based laptop ever being "produced on a massive, industrial scale"
My counterpoint: it's revolutionary because it's the first ARM-based laptop which doesn't have to make a ton of compromises to make it out the door.
It's got mass-market appeal ("widely popular", unlike the PineBook Pro).
It's not limited in functional scope ("fully functional", unlike Chromebooks).
It's not limited in computing power ("powerful", unlike any other ARM laptop).
It's not restricted to a subset of tasks ("general-purpose", unlike the iPad Pro when treated as a laptop).
It's being mass-produced for retail, i.e. it's not a limited-run or a prototype ("produced on a massive, industrial scale").
Every other attempt at an ARM laptop has made one or more of these compromises; the M1 Macbooks (and the M1 Mini) don't have any of these compromises, meaning that it's fit-for-purpose for the vast majority of laptop users (those for whom a Macbook would have sufficed before the M1 line).
I think being good and useful is pretty revolutionary; I haven't seen another ARM laptop offer that, and those are pretty important features.
This should not be understated. You could hand an M1 Mac to anyone and unless they were technical and understood what you were handing them, they would neither know nor need to know that it has an entirely different CPU architecture.
It's totally transparent. Not one application did I try did it do anything other than just run as I would have expected. Heck the vast majority of x86 Windows software I tried - either under Crossover (commercially supported version of WINE) or in the Windows 10 ARM beta under Parallels - worked just fine.
And very speedily too. It was not apparent that emulation was going on - at all.
It really is quite astonishing. As they say, seeing is believing - in this case using is believing. The overall feel of the system is just not something that's easily to articulate. It's not just about running a benchmark or application quickly; the whole thing is just more responsive.
The only reason I took my M1 Macbook Air back on the last day of the return window was it turns out to do everything I want if it just had more RAM. Which was NOT my starting position; the M1 ran so well that I ended up realizing it could do everything I wanted and more if I could get at least 32GB of RAM - 64GB would be perfect.
So I'm rather impatiently awaiting the next round. As soon as I can get more RAM I'm so getting another Apple Silicon laptop!
It's night and day. Readily noticeable by non-technical people. It's not just faster, it's snappier, more responsive and just has an all around different feel.
The writing has been on the wall for almost a decade, but the execution required to achieve it is definitely remarkable.
My M1 MBA is better than my 2020 Intel MBP (which cost over 2x) in almost all ways that matter: quieter, cooler, battery life, and incredible responsiveness.
And x86 will never catch up. That's why it's revolutionary - it has destabilized the existing regime.
True, AMD has a few tricks up their sleeve, and is a node behind Apple at TSMC, but on a purely technical level it's clear that Apple ahead of both AMD and Intel.
Now, power consumption may be the end all be all for some people. If your work tasks actually don't require a lot of computational power, then I can see it being much more valuable to have the flexibility of a cool laptop with a long battery life.
But, touting the M1 as revolutionary, because all that matters is power consumption, feels like a disconnect to me. I'm all for people liking stuff they like. But, I don't get what's revolutionary at all. I was interesting in hearing what people thought about it though.
> in almost all ways that matter: quieter, cooler, battery life, and incredible responsiveness.
This discussion also comes up again and again on HN on the M1. People say weird things like that the never experienced such a jump in computational power from one generation to the next. And what they tend to mean is from older apple hardware to newer hardware.
Before releasing it though, they worked up to it (in secret), 2.54 centimeters by 2.54 centimeters.
To have that freedom in a 3 lb laptop with virtually no performance compromises is incredible.
A revolution doesn't have to benefit you. This benefits Apple financially, any benefits you receive are secondary.
https://www.cpubenchmark.net/singleThread.html
(Available in a fanless ~$1k laptop, no less!)
Similarly on the Intel side they have the 11900 ranked higher than the 11900K. Which again, there's no way that's true since the 11900K is literally an 11900 with a higher power limit & higher turbo frequency. And then the 11900KF, which is just an 11900K but with the iGPU disabled, is then somehow a lot faster than both?
Actual binnings & achievable frequencies vary, of course, but these orderings aren't passing a sniff test, either.
https://www.anandtech.com/bench/product/2687?vs=2673, https://www.anandtech.com/bench/product/2687?vs=2637, and https://www.anandtech.com/bench/product/2687?vs=2787 are better comparison points here. The M1's single-core performance is of course very strong, but I'm not sure why you're surprised that it's "still" the fastest single-core CPU since these CPUs are all the same generation. There hasn't really been anything newer.
In any event, you'll notice M1 achieves an extremely competitive level of performance with a peak power of 22% of that used by the Ryzen 5950X and 10.5% of the 11900K. That's crazy!
In terms of efficiency there really is no comparison the M1 is a remarkable product.
We'll have to see how the M1X/M2 goes, but given what we've seen so far, we should see some stellar results from an M series chip designed for a workstation class laptop.
We'll see how they scale up, with the new MacBook Pros.
That isn't impressive, no. Power scales non-linearly with clock speeds (and thus performance across a particular CPU design). Compare a 3700x vs. a 3800x for example: https://www.anandtech.com/bench/product/2665?vs=2613 - Huge increase in power draw, barely any performance increase.
You'd have to compare at equivalent power levels to draw a meaningful conclusion like that. Especially since the 5950X is paying power for things the M1 just doesn't have at all, like the large amount of PCI-E lanes. But otherwise for a wall-powered system quadrupling the power for doubling the performance is a quite straightforward "yes please!" tradeoff to be made. Often the gains are much smaller than that.
Of course the M1 also isn't designed to excel at heavily multithreaded workloads having "only" 4 big CPU cores, so it's an unfair fight in that direction as well.
Macs suddenly switched to these ARM processors. That's the revolution, and the number is M1, but it's based on the gradually developed A series.
The m1's compute/watt will enable designers to change the dimensions of a laptop to be even smaller and compact. If the iphone's a15 starts to be near m1 compute-power, we could actually see the iPhone merge to become the device powering our monitors and keyboards. The new possibilities of such a revolutionary technology is endless!
I highly doubt that because it means Apple will sell zero macbooks and they don't want less sales.
I don't think it's arguable that the M1 wasn't a game changer though. M1 firmly established Apple's design prowess at the laptop/desktop class chip game. It was no longer arguable that mere phone chips weren't comparable to desktop class CPUs. That claim is dead. It also enabled performance and power efficiency levels years ahead of the competition, giving tangible immediate benefits to customers.
So revolutionary? It's a matter of opinion, but I think it was a watershed moment and permanently changed the technical landscape for CPUs running 'full blown' desktop class software.
It sounds like you are conflating AArch64 with Apple. Apple had nothing to do with AArch64, it was developed by ARM in-house; Apple simply released the first AArch64-compatible core (outside of ARM's Cortex series).
Then AppliedMicro had a hardware version to demo.
Apple simply had the first to market consumer product with one. Them buying up a good chunk of the fab space needed to produce them at the time probably contributed to that.
Apple is able to do what they want because they have an architectural license; not a core license. It's the same reason Nvidia, Cavium, etc are able to design their own cores irrespective of the Cortex series. That's what an architectural license is for. While MediaTek, HiSilicon, etc are limited to using the Cortex SIP.
It's weird to look at this in the context of Google and T-Mobile almost shipping the first Android devices with a $10 a month unlimited data only plan...
Riding Wright's Law.
> I'd argue the M1 is a revolutionary product on the same level as the first iPhone.
This was first obvious with Apple's move to 64-bit for mobile. And should be more obvious with the Apple Watch. While M1 is quite the competitive advantage, there's nothing on the horizon to compete with Apple's wearables; Apple dominates both the profits and marketshare.
I don't know what superlatives to use. Nor do I know how to summarize, much less describe, Apple's strategy. But it's working.
Apple focused on profitability. Capture the highest margin market segments, then ride the price curve down. This sucked all the oxygen out of the competition.
Similarly, Apple's monospony strategy repeatedly boxed out competitors.
Most infuriatingly, Apple laser focused on select differentiators and competitive advantages to the exclusion of many others. For example, thinness for thinness sake. At the expense of reliability and repairability. Why? Because they could and copying them was prohibitively expensive. Again, boxing out competitors.
--
Contrast and comparison sometimes helps understanding.
Two other examples of Apple's Druckeresque strategy are Tesla and SpaceX. Capture the profits, plow the capital back into R&D, relentlessly drive Wright's Law.
My hunch is the core strategic focus is on lowest cost of capital. What moat most protects the dominance of AWS, Apple, Tesla, SpaceX? I think, compared to their competitors, it's their lower cost of capital?
Duh, right?
So while I think it's fine to speculate and cheer. But it's hard to get excited by competitors who a) don't have cheaper capital and b) aren't driving Wright's law for crucial competitive technical advantage.
Toyota used to have these advantages. But somehow lost the narrative. I think the case studies will argue that Toyota missed two opportunities. First, they punted on owning their own digital hardware and software, whereas Tesla treated the car as an iPhone on wheels, a new platform. Second, most importantly, Tesla jumped into EVs at the very moment Li-ion became feasible, and decided to relentlessly drive Wright's Law for battery tech. The other manufacturers simply waited too long to jump in. Tesla's Li-ion flywheel (access to capital, scale, Wright's Law) will kick in within a few years, cementing their dominance.
Of course Tesla's story isn't finished. Like all hard driving efforts, balanced on a knife edge, Telsa could implode. But if they avoid a cash crunch, it's hard to imagine any EV competitor catching up. In that way, Apple's access to cheap capital is great deal more robust.
FWIW, I anticipate Starlink will become SpaceX's Apple style money printing machine. And once that revenue kicks in, the existing cell and cable carriers will be buried. The Mars mission and national defense parts of SpaceX's story are fun and legit. But unlike all their competitors, SpaceX will be self-sustaining on Starlink.
--
Thanks for reading this far. I write to understand. And this narrative has been eluding me.
--
PS- The outlines of next technical turf battles are pretty obvious.
Apple's next big challenge is baseband chips. They know this, of course. But for whatever reason, eliminating their dependence on Qualcomm is very hard. And whoever productizes the successor to 5G, whatever the satellite uplink chipset shapes up to be, will be sitting very pretty.
The Li-ion story wrt BEVs now seems pretty obvious. Everything from light trucks to skateboards will be almost fully BEV by around 2030. Tesla will produce about 1/10th of the world's batteries; and maybe a 1/10th of the Model 3 sized automobiles. And like Apple, they'll capture most of the profits.
The next frontier will be hydrogen. The hydrogen story today looks like Li-ion around 2005, +/- 3 years. Hydrogen's first use cases will be all the stuff BEV missed. Like trucks, transportation, utility scale energy storage. Someone like Toyota could survive long enough to see their hydrogen bet pay off. But I think it's more likely new entrants will access to cheap capital will prevail.
Any way.
Quite the contrary. Apple focuses on experience - and knows that if you deliver a kick ass experience, profit will naturally follow.
Apple also is one of the few tech companies that seems to be willing to say "no" and not be all things to all people. It's often misinterpreted as only targeting the profitable markets, as if that's a bad thing. Not pursuing unprofitable markets seems to be common sense to me - call me crazy!
Microsoft of the 80's and 90's focused on profitability and chasing relentless backwards compatibility; and it worked - for a while. Until you had the Windows division actively undermining other portions of the organization it saw as competition. Memes like this are no accident: https://www.globalnerdy.com/2011/07/03/org-charts-of-the-big...
Then the next wave of computing - mobile happened and they suddenly were on the outside looking in. Much like the likes of DEC, Prime and heck even SUN were when the PC revolution happened and dethroned mini's.
As for baseband chips - radio is freakishly unforgiving. Apple did buy Intels baseband modem operation lock stock and barrel, so in three to five years - who knows? They certainly are motivated and have the resources to make a solid go of it. And they have certainly proven they can expand their expertise rather competently when they are motivated to. I don't think the Swiss are as dismissive of the Apple watch as they initially were; some even admitting the band designs Apple came up with were an improvement. That you can take segments in and out of the metal segmented band with just your thumbnail is, dare I risk saying it, revolutionary. If you don't think so I have a Timex band the flummoxed experienced jewelers - lets see how you do with it.
Here you have undoubtedly some of the finest processor technology available, manufactured in the hundreds of millions, wasted by the constraints of its platform. If the Pi foundation had access to cores, processes, and even pieces of this level of technology the overall computing world would be so much more capable.
Even beyond that, the iPhones themselves? Severely limited by the I/O they possess. As these devices age they could have several recycled general computing purposes but they're constrained by the single lighting port. iPhone use NVMe storage, they have PCIe lanes!
All this work, effort and engineering for a platform who simply seeks to make more off of the revenue of whales in their app stores.
Edit: I think my comment around Pis is a little misleading. What I meant is the general educational Linux community. Not specifically the Pi foundation. If after X years Apple unlocked the bootloader and you could install other operating systems to better leverage the hardware I'd feel better about the state of it.
I mean, ignoring that there are some fairly serious number crunching things to do on iPhones (I think the lidar and basic 3D scanning features intro'd with the iPhone 12 Pro are already quite useful professionally for example, like cameras themselves on smartphones), how can you say this following Apple's switch of Macs to their chips too? Obviously the economies of scale and constraints generated by iPhone development has fed Apple's move up the stack in classic disruptive fashion to the iPad and then last year the Mac. Hwever you feel about smartphones themselves, it seems a little more than a little odd to see this comment now in 2021 given that it's going straight into more classic personal computers as well?
It looks more like sour grapes that one of the most prestigious computing companies on the planet is out-engineering the "foundation" spin-off of Broadcom, created (very successfully, I might add) to breathe life into an aging chip foundry.
He wants that chip in a raspberri pi, not sure what you call it, it's the quite the opposite of sour grapes
Seems appropriate to me.
Everyone is buying a magical teleporter but it'll only take you down to the corner store. Doesn't matter if 95% would only use it for that anyway, the fact the other 5% can't is what makes me sad.
Depends I guess? Having a lot of power on tap can show up in transparent ways as well as obvious apps. Apple is definitely leaning heavily on their SoC for a lot of their computational photography stuff. Pros may prefer to just shoot RAW of course for decent reason, but having a system make "good shots" fairly point-and-shoot at the cost of a ton of compute is a real value add for lots of regular people. A lot of people also do some real gaming on their iPhones these days, and those can absolutely push the GPU. Maybe that doesn't fall into your definition of serious, but I don't think it's quite fair to dismiss any use-case valued by customers. There are also extremely practical energy saving issues like the race to sleep. Often the faster a chip can do a job and then return to hibernation, the better the energy efficiency. A lot of normal usage is very bursty, and like it or not on the modern web there are tons of sites that can hammer SoCs pretty hard. Everyone cares to some degree about battery life and responsiveness in handhelds.
If there was some big cost for this that'd be one thing, but there really isn't given the economics of modern silicon design and fabrication. Apple amortizes R&D big time across the iPhone, iPad, Mac, and even stuff like the Apple TV. Relentlessly pushing forward the units on the phones feeds directly into everything else. And even if most people only use the full potential a fraction of the time that may be a very valuable fraction, and what will take off in the future that might use it isn't always clear.
Though one thing that is clear is that wearable displays and serious AR/VR are the big next disruption/extension event for computing, and Apple like every other player needs to be ready. That's going to take a ton of compute power along with highly evolved environmental sensor usage and fusion. They have to be working towards that for years and years beforehand, and clearly are.
The article mentions that several leads from the Apple chip division bailed for Nuvia, so at least the future of non-Apple boards looks a bit brighter.
Even if they had access to it I don't think it would be anywhere near as affordable as the Pi foundation wants their devices to be. An M1-powered Raspberry Pi would be amazing, but also probably several hundred dollars.
My iPhone 11 and iPad Pro have more, and more varied, I/O devices available than any of the ~6 computers and dozen or so video game consoles in my house. Like, I could buy more for the others to make them compete, but the i-devices just come with tons of stuff, plus can connect to lots of the same USB devices my other computers can (including, notably, anything MIDI), plus most of my bluetooth peripherals. And they can use external monitors/TVs (mirroring only, but still).
I guess there's a chance my Nintendo Switch comes close, but then, it can't connect to as many BT and USB devices.
Last year's M1/A14 cores had performance gains due to the combined architecture improvement and the shifting from TSMC 7nm to 5nm fabrication. At the time it was difficult to determine the % of gains attributable to either, so now it seems we have a better idea that node advantage could be the majority.
It'll be very interesting to see what happens in the next M-series chip announcement, e.g. M1X or M2. I still think it's plausible Apple switches the M-series chips to 3nm first in order to cement their superiority over Intel. That said, porting architectures across nodes is no small feat (as we saw in Intel's Rocket Lake, though that was small -> big).
I've been waiting for years for someone other than Apple to make a fully custom ARM SoC that can compete with Apple.
I suspect there isn't a lot of incentive for Qualcomm to do it as they as nearly a monopoly in the Android space.
I've been hoping that AMD or Intel would step up and make an ARM core that crushes Qualcomm.
I just got a used iPhone 7 for my so and despite being 5 years old is still really smooth running the latest iOS 15. It's the only iPhone I own (the rest are Android) and I'm a bit jealous.
But the end result is nobody really wants to compete against Qualcomm in the mobile phone SoC space, since Qualcomm's radios means they win by default. Then in the laptop & server space, Intel & AMD have no incentive to switch to ARM instead of sticking with x86. At least not at this point in time.
that is nonsense. I switched from iPhone to a Pixel 4 post-CSAM scandal and the difference is night and day. My phone now makes me wait. Every single fucking day I wait. Wait Wait Wait. Slow Slow Slow. And this is a phone without the bloatware. Fucking miserably slow. Not in absolute terms necessarily, it would have been competitive 10 years ago. But in 2021, Qualcomm is fucking over every single Android user and their management should be ashamed.
I assume nonsense about single core performance being good enough is how the industry (Apple aside) has ended up in such a sorry state.
For carriers this lets them get out that first 5G phone and all that nonsense. For OEMs this drastically reduces engineering complexity (and therefore improves margins) by avoiding a separate radio chip.
I've been an Android user since the original G1, and every year I wonder why I don't switch to an iphone. Maybe this year.
I'll probably suck it up and buy a pixel 6 once graphene is available for it.
Also, I don't think it's fair to assume that Graphene is not playing some role in the slowness (though I don't believe most of the currently-implemented hardening is expected to create dramatic slowdown). Perhaps you should try running the stock ROM for a day or two (with the Google apps disabled in Settings, of course).
Mutually exclusive statements.
That may change soon! Samsung and MediaTek are hot on their tail.
https://www.notebookcheck.net/MediaTek-Dimensity-2000-to-fea...
Same for media codecs, and lesser extent, NPU kind of stuff.
>Comparative subsets would 5950X 7.29 int / 9.79 fp, 11900K 6.61 int / 9.58 fp. versus 7.28 / 10.15 on A15
A510 is the next best thing. Like bulldozer, it will have two integer units sharing a single floating point unit. I doubt companies will ship 8 A510 cores which means that overall little core performance will likely not be improving a huge amount.
In fact, it's looking to me like A79 is going to be the "little" core while X2 will be the big core and a couple A510s will be strictly limited to secure low-level processes.
macOS/iOS vs Windows or Linux.
Usain Bolt is the fastest person on earth. But if he’s wearing 5 kilogram shoes, he’s not going to be any faster than a normal person.
On a very practical scale; none. It'd make some nerds happy but it's like a subset of a subset of a niche population.
> The framework and librem laptops show that there is a market of people willing to pay for upgradable devices which are independent from makers and free from out-of-the-box backdoors.
All it shows is that there is some market for it but I think you're either severely overestimating framework and librem's scale or are severely underestimating intel and AMD's. I'm extremely skeptical that either company cares about that niche nor do I think they'd rework all their processors just to satisfy that niche.
Too small to measure: the subset of privacy activists who are concerned about it is orders of magnitude smaller than the number of IT workers who use those features, and that's a tiny fraction of the number of people buying computers.
Calling it as a backdoor is your choice but consider that using a term not following its standard definition is not likely to be effective or good for your credibility if you're trying to draw attention to something more important in the future.
IME features maybe useful for some people. But there are people who don't want it. I don't think it would be expensive to make it possible to disable it via eFuses.
Remember that linux was once much smaller than any commercial UNIX. So was Apache and Nginx. Linux is still very unpopular among gamers; that didn't prevented Valve from investing in what now is their lifeboat.
A small market should not be seen as an excuse to not invest in something that will certainly bring goodwill.