<i>What would happen to the cost, battery life, and size of an A10-powered MacBook Air?</i><p>It would be worse in almost every aspect.<p>The cost wouldn't change much, and Apple wouldn't profit much more from the switch. The cited CPU cost is the <i>suggested retail price</i>. Apple's volume lets them negotiate huge discounts.<p>Battery life would go up a bit, but battery life is already pretty good on Apple laptops. My 2013 11" Air gets 8 hours easily, but can go as long as 12 or as short as 2 depending on screen brightness and CPU usage. Unless you're keeping the CPU busy, the screen is the biggest consumer of power. Cutting CPU power consumption in half would only increase battery life by 20-ish%.<p>The PowerPC to Intel transition worked for several reasons. Most importantly, Intel CPUs were much faster than PowerPC. In everyday usage, the fastest ARM chips are 10x slower than a quad-core Haswell. Moving to x86 also increased the features of Macs, allowing them to dual-boot Windows and efficiently virtualize other x86-based OSes. Switching to ARM would backtrack on both fronts. An ARM-based Apple laptop would have no boot camp, no Windows virtualization, and no efficient emulation of legacy applications.<p>CPUs are only a small part of why tablets have longer battery life than laptops. Tablets have no keyboard, trackpad, or hinge, so they can basically be a giant battery with a screen attached. The iPad has a bigger battery than the 11" MacBook Air, despite the Air weighing 50% more and taking up 30% more volume. (Edit: This has recently changed. The 11" Air has a 38 watt-hour battery. The iPad 3 and 4 had a 42 watt-hour battery, but the iPad Air has a 32 watt-hour battery. Still, it's even smaller than the earlier iPads, massing less than half the 11" Air.)<p>In short, it doesn't seem worthwhile to do all this work and sacrifice so much performance for some incremental increases in battery life and profit.
"The aging x86 architecture is beset by layers of architectural silt accreted from a succession of additions to the instruction set... Because of this excess baggage, an x86 chip needs more transistors than its ARM-based equivalent"<p>I wish people would stop saying this. On the inside, x86 CPUs are basically RISC. There is a translation layer from the publicly facing instruction layer to the internal representation. The transistor and power budget for this translation layer is absolutely trivial.<p>Intel has an enormous amount of expertise in the actual manufacturing and design layers, as evidenced by the rate of improvement in their Atom CPUs (striking range of ARM, actually), integrated GPUs, and quasi-GPU compute cards. They are not in danger of "losing" in the long term in a performance or performance / watt race. The risk is they get disrupted due to all CPUs turning into a commodity in roughly the same way that RAM is a commodity.
From the article:<p><pre><code> > Because of this excess baggage, an x86 chip needs more
> transistors than its ARM-based equivalent, and thus it
> consumes more power and must dissipate more heat.
</code></pre>
This is true but it ignores the primary reality of "desktop class" processor design today: <i>RAM is the bottleneck in a really major way and most of a desktop class CPU's transistors are dedicated to overcoming this.</i><p>In the ancient days, CPUs ran synchronously (or close to it) with main memory. Hasn't been that way for decades. CPU performance has ramped up so much more quickly than that of main memory that it's ridiculous.<p>And this is where most of your transistors are spent these days - finding ways to allow the CPU to do some useful work while it sits around waiting for main memory. Look at a modern i5 CPU die:<p><a href="https://www.google.com/search?q=intel+core+i5" rel="nofollow">https://www.google.com/search?q=intel+core+i5</a><p>Things to note:
- Tons of L1/L2/L3 cache so we can keep things in fast memory. The transistors dedicated to cache <i>dwarf</i> those allocated to the actual processing cores, let alone the parts of those processing cores dedicated to those crufty ol' x86 instructions
- Lots of transistors dedicated to branch prediction and speculative execution so we can execute instructions before we've even waited around for the data those instructions depend upon to arrive from slow-ass main memory<p>Sure, mobile ARM chips are tiny and efficient! They run at 1-2GHZ while paired with fast RAM that's not <i>that much</i> slower than their CPUs. They don't need to devote gobs and gobs of transistors to speculative execution and branch prediction and cache.<p>But all that changes if you want to scale an ARM chip up to perform like a "desktop-class" Intel chip. You want to add cores and execution units? If you want to keep them fed with data and instructions you're going to need all that extra transistor-heavy baggage and guess what -- now you're just barely more efficient than Intel, <i>and</i> you can't match Intel's superior process technology that's been at least a transistor shrink or two ahead of the competition since the dawn of the semiconductor industry.<p>Eventually, yes, the ARM chip makers will solve this. RAM will get faster and processes will be enshrinkified. Just understand that transistor size and pokey RAM are the bottlenecks, not that nasty old x86 instruction set.
I think it's funny that the main reason people cite the need to switch from Intel to ARM is power consumption. A big reason to stay on Intel is that a lot of people have real needs to run Windows software (whether in Bootcamp or a VM). Switchers feel a lot better jumping to the Mac if they could fall back to Windows if they wanted to.<p>Regarding battery life: The latest MacBooks get 9-12 hours of battery life. I haven't experienced battery anxiety on a Mac in a long time. In contrast, my iPhone is dead by the end of the day and watching it creep below 50% makes me start thinking about the nearest Lightning adapter (yes, I understand the Mac has a <i>much</i> larger battery and cell radios are power hungry).<p>Put another way, max power consumption on an iPad Air is ~11W[1]. The max power draw on a Haswell MacBook Air is 15-25W (~50% improvement in battery life from 2012 to 2013, which had 21-34W max draw)[2][3]. Given that Macs have more space available for batteries due to larger screens and the need for keyboard and trackpad, I don't see power consumption argument holding water.<p>[1]: <a href="http://www.anandtech.com/show/7460/apple-ipad-air-review/3" rel="nofollow">http://www.anandtech.com/show/7460/apple-ipad-air-review/3</a><p>[2]: <a href="http://www.anandtech.com/show/6063/macbook-air-13inch-mid-2012-review/7" rel="nofollow">http://www.anandtech.com/show/6063/macbook-air-13inch-mid-20...</a><p>[3]: <a href="http://www.anandtech.com/show/7180/apple-macbook-air-11-2013-review/2" rel="nofollow">http://www.anandtech.com/show/7180/apple-macbook-air-11-2013...</a>
The real weakness of Intel in power consumption relative to ARM hasn't been so much in the chips as the chip sets, software and the balance-of-system.<p>To create a low power system, ALL components need to be low power, and it is easier to start from a system built to be low power and scale up capabilities rather than go the other way around.<p>For instance, I have a Windows-based laptop which is a great machine, but if I have a web browser open, any web browser, the fan spins, it gets hot, and battery life is less than 1.5 hours.<p>Is it the fault of Windows, the browser vendors, the web platform, adware, crapware, who knows what? It doesn't matter, but controlling power consumption on a legacy platform is a game of whac-a-mole that doesn't end.<p>Because Windows users expect to plug in devices that draw power from USB, a Windows tablet has to have a huge power transistor to regulate voltage, a power supply system scaled up so it can supply enough power through the USB port to charge an Android tablet, at this point you might add the fan and then you are doomed.
Apple does look like it's making chips to handle heavy workloads, not just to compete with the latest Krait or whatever. But I'm not sure ARM Macs are the direction they'll take that.<p>It'd make some business sense for them to instead position iOS so it can take over more and more traditional Mac duties. The IBM push could be an example of that. Investing in iOS gives Apple the tight control and the 30% cut they're used to from that space, and it avoids the Windows-RT-ish heartbreak of "why is this ARM OS like my Intel OS but without all my apps?" (If there <i>were</i> Mac-on-ARM, I'd expect it to be Mac App Store only.)<p>Anyway, the A7 is already a beast (<a href="http://www.anandtech.com/show/7910/apples-cyclone-microarchitecture-detailed" rel="nofollow">http://www.anandtech.com/show/7910/apples-cyclone-microarchi...</a> does various measurements, <a href="http://cryptomaths.com/2014/04/29/benchmarking-symmetric-crypto-on-the-apple-a7/" rel="nofollow">http://cryptomaths.com/2014/04/29/benchmarking-symmetric-cry...</a> is an interesting case study), and there are still future process nodes and microarchitecture changes that will let them make better chips. I don't know if an ARM MacBook Air is specifically where this goes, but they're certainly making ARM capable of more serious stuff.
I'm saddened that nobody ever brings up video games when talking about these architectural shifts. Practically all PC games in the past few decades have been written for x86, and most of them will never be patched for ARM compatibility. These games are as important to many of us as movies or music, and yet I fear they're destined to disappear from our cultural memory if this shift ever happens. Virtualization just won't cut it; most modern games barely even run with Wine, to say nothing of performance. And given the slowing of Moore's law, we can no longer count on emulation to give us seamless reproductions a few years down the line. Does nobody care? Why doesn't anyone say anything? I love my Mac, but if I have to choose between ARM and my Steam library, I'll choose Steam and begrudgingly go back to whatever Windows version still supports it.<p>On a related note, I think it's important to differentiate between utility software that's assumed to be temporary, and one-off pieces of software that are intended to live forever. I wish there was an easier way to write software in such a way that it can easily be guaranteed to run in the future, no matter the architecture. (Open source is not a guarantee. Ever try compiling the source to a AAA game?)
If this happens I hope entry level Mac’s get below $500. I bought a MacBook Air for use as a Windows machine with the plan to learn iOS development at some point. It was worth the cost since I can use it as my day to day machine running Windows. I wouldn’t be able to justify spending $1000 on a Mac Laptop that could not run Windows. I could justify $400 for an entry level Mac Mini running ARM but not a lot more then that.
These arguments make no sense. Aside from anything else, the computing power of the Mac I'm currently using—quad 2.4GHz Ivy Bridge—is so far in excess of anything that is available in the ARM architecture that it's difficult to see this being the case at any point.
I'm probably in the vast minority, but I tend to use apple hardware to run windows... so I'm hoping this doesn't come to pass. I use OSX occasionally, but the thing that got me to switch to apple in the first place was bootcamp.
Losing native Windows virtualization seems like a big deal. That was a huge selling point with the switch to Intel. It's become indispensable for many Mac users and was the reason many Windows users were able to switch.<p>Unless Microsoft also drops Intel, I don't see this happening.
The "transitions" comments leave out a big piece of functionality; certainly big for me, and I imagine big for others. That's VMs for Windows and Linux. I have all kinds of Linux VMs I run on my Mac, and I imagine others that need "that one Windows app" run Windows a lot in VMWare and VirtualBox. Going to ARM would torch a big part of Mac functionality for me.
From what i've understood from benchmarks and reviews the latest Intel atom processors have a better power/performance ratio then arm processors. I'm not sure if the performance/price ratio is better though, does anyone know?
If Apple were to switch from Intel, I'd probably have to (reluctantly) go back to a Windows laptop. I love my MacBook, but most of my money is still earned working with clients who use Windows environments. There's still some software that never made the jump to Mac either, for which I still have to use Parallels Desktop. The best bit of having a Mac is that I can run OS X, Windows & Linux all on the same box.<p>Of course, if Apple is also making their own x86 compatible chips, that's a different story. I don't need an Intel chip specifically, I just need something that runs Windows / x86 perfectly....
As much as i want this to happen. I dont see this coming in the near future.
Why would they release the new MacPro with Intel Xeon if they had ever planned to switch away? And would MacPro be staying in x86 land if Apple decide to switch the rest to ARM?<p>Another obstacle is Thunderbolt. This DisplayPort + External PCI-Express Cable is totally in control by Intel. AMD's version is based on USB 3, which is an ugly hack that Apple will unlikely use.<p>The performance gap between Haswell and A7 is huge. Watts to Watts, at Notebook / Desktop power range Intel wins hands down. Although the gap is shrinking with each Ax SoC.<p>Then there is the part about Intel Atom losing on performance. Which is wrong. Intel Atom SoC performs really well. It didn't get much wins simply because of its ecosystem and prices.<p>The Mobile SoC market are operating at thin margin comparatively speaking. Even if Intel are offering Atom at the same price, why would any OEM wants to be bound by Intel & x86 again? So Intel decide to offer those SoC at a lost and what happened? On the Western world it is dominated by Qualcomm where it offer better solution or cheaper TC with its integrated Modem. In Eastern World or China it is hit by "8 Core" marketing team from MediaTek. Everyone thoughts more Core = better.<p>I dont see how Intel is going win this Mobile battle. Apple will pretty much drive Intel to where they want, which is to Fab SoC for them.
I blogged about this in April. <a href="http://vishaldoshi.me/2014/04/25/apple-intc/" rel="nofollow">http://vishaldoshi.me/2014/04/25/apple-intc/</a><p>It looks like Chromebooks are a very popular format and I think an ARM based 'AirBook' could compete in that space.<p>Napkin Math<p>MacBook Air current generation (mid-2013) retail price: $999 ($1099 for 13″ model with same CPU).<p>Intel Core i5-4250U, Tray: $315, <a href="http://ark.intel.com/products/75028/" rel="nofollow">http://ark.intel.com/products/75028/</a>
(sure Apple will be getting large discounts on this, but it can't be that large, since $INTC has ~60% gross margin overall)<p>Apple A7, Tray: $20 (estimated)<p>Intel Atom E3827, Tray: $41<p>Tray = 1000 pcs;<p>The Core i5 has a 15W TDP; 1.3 Ghz clock (turbo to 2.6Ghz); 2 cores, 4 threads. Sunspider 250ms.<p>The Apple A7 has a 2W TDP; 1.3Ghz clock; 2 cores, 2 threads. Sunspider 397 ms.<p><a href="http://cpuboss.com/cpus/Intel-Core-i5-4250U-vs-Apple-A7" rel="nofollow">http://cpuboss.com/cpus/Intel-Core-i5-4250U-vs-Apple-A7</a><p>It’s not looking all that different! Esp. when you take into account that the A8 will be twice as fast (think Tegra K1) – i.e – a Sunspider score of 200ms maybe?
Had a hard time loading it, so here's the cached version: <a href="http://webcache.googleusercontent.com/search?q=cache:SDIaXc5Z72sJ:www.mondaynote.com/2014/08/03/macintel-the-end-is-nigh/" rel="nofollow">http://webcache.googleusercontent.com/search?q=cache:SDIaXc5...</a>
The Windows kernel guys were evaluating ARM systems for us, and the gist was "Run away, don't walk; the memory systems on those things are terrible."<p>I like the ARM architecture a lot. It's simple and easy to write software for, all the way from embedded controller stuff to "real" operating systems. But they're not all that great at doing massive computation. We were going to use one in the video path of a popular gaming system for a while, and it turned out to be inadequate by at least an order of magnitude (probably a factor of 100, but comparing CPU cycles to GPU cycles is pretty unfair).<p>Intel is executing really well, and it'll probably take an alien invasion to dethrone them.
Apple needs x86 to run existing apps. Who would accept a a big slow down for a slightly cheaper laptop. x86 intel chips are a RISC design with a glue layer which grafts x86 on the outside. Could Apple design an x86 chip with an ARM RISC core hidden inside? Sounds pretty complicated, I doubt they have a big enough hardware engineering team. If there was an ARM core inside could it be exposed for recompiled software to take advantage of it? Big issues managing state/contention between 2 different ISAs?
This is a horribly low-quality article. The central proposition is <i>entirely analyst speculation</i>, with no actual information to hook it onto that doesn't date back years. The author appears, from how he gets details subtly wrong, not to actually know anything about the history of CPUs (and not to have bothered e.g. checking Wikipedia).<p>There is nothing here that is backed-up news whatsoever.
According to Geekbench the current CPU in the iPad Air (the A7) scores about the same as a Core 2 Duo from 2006 (the E6600). Nothing to sneeze at and definitely getting close to the low power Haswell CPU's in systems like the Macbook Air. But I reckon it'll be a few generations of A series CPU's before they reach the point where they can challenge Intel for the crown
For someone who's not very knowledgeable about architecture shifts, what software changes would be require to make this happen? I'd assume that the Mac OSX itself might be able to make the shift easily, given that Swift and ObjC code seems to run on both iOS/ARM devices and MacOS/x386 machines. What else will have to change? Or is this purely a hardware choice?
<i>If you wish to make an apple pie from scratch, you must first invent the universe</i><p>How far does Apple want to go? The rabbit hole is pretty much infinite. Do they want to make their own plastics and paints and metals? Do they want to mine the materials for all of the above? Do they want to make the mining machines for their mines? Etc, etc, etc.
An ARM based Air might make a nice facebook machine but what about Pros/Desktop and the high performance market? I really doubt there are any ARM chips anywhere near the performance of an i7. What are they going to do segment their PC lineup or just abandon the high performance side? I doubt they are willing to do either.
While I think the article's prediction is incorrect (see my other comment for why), it did give me an idea: Why doesn't Apple design their own x86 CPU?<p>Their work on the Ax series has probably taught them quite a bit that could be ported over to x86-land. Also, Apple could leave out x86 cruft they don't use: legacy addressing modes, PAE, etc. And of course, they could design the CPU specifically for their products instead of searching for the closest match sold by existing vendors (or cajoling Intel to tweak their designs).<p>Apple already has strong relationships with fab companies. They have the talent and teams to design such a CPU. One wonders if they're already working on such a thing. Even if it never shipped, it could be used to negotiate lower prices from Intel.
What seems more likely to me is a hybrid ARM / x86 machine that switches dynamically between processors, kind of like how some macs can switch dynamically between their integrated and discrete GPUs.
apple has been known in the last decade for picking the best hardware for the job. i don't see how ARM chips can compete with intel chips in terms of performace or performance per watt - unless you want a 4W macbook. (hint: you probably don't.) ARM competes very well in performance per dollar - but AMD does, too, and I don't see macs with AMD CPUs in them.<p>i can see a laptop-sized ARM-based product from apple, but it won't replace any of the macs, it'll be something completely new - let's call it a macpad.
> Googling “Mac running on ARM” gets you close to 10M results.<p>Why should we listen to someone who doesn't know what quotes do in search queries?
The transition from powerpc to x86 was surprisingly relatively painless. There was some pain but most apps ran fine in the translation layer and eventually the native apps moved over.<p>It'll probably be even easier the next time since apple has done it once already and knows how to provide the proper dev tools.
this is about the most poorly written article i've ever read! well, second that to some of the trash on techcrunch. how about some benchmarks of compiling code on intel vs arm instead of saying that since ipads (arm based) cost more than macbook airs (intel), so therefore intel will fade from apple's line of laptop/desktops. i think arm processor laptops are going to become mainstream, but not for most of the arguments specified in the article. i believe this fate is still far away as x86 still offers total raw compute power over arm, even though arms are more energy efficient.<p>also, the self noted digressions in the article arent even funny, feels like someone with an english degree and subscription to "i can spell x86" magazine wrote this article