Hacker News Reader

You say that ARM chips are 10x slower than Haswells, but the difference isn't actually that large anymore and for Apple's A7 it's around 3x versus a normal desktop chip like the 4770 and only 1.7X versus the 4250U you'd find in a Macbook Air[1]. Diminishing returns is a constant factor in engineering and I'd expect it to be very hard for Apple to make up the remaining distance between themselves and Intel, but we shouldn't exaggerate how large that difference is. And it is possible that some hypothetical A8 could be made competitive with Intel in the 15W range due to not having to make the compromises required to hit 4 GHz.

I do agree that the ~10W you'd save by using an A7 versus an 4250U in a Mackbook Air would be a false economy.

EDIT: To explain that bit about targeted power ranges a bit more, to hit high frequencies at a given process node (like 45 nm or whatever) you need to break up your logic into a fairly high number of simple stages separated by clock driven latches. A core that doesn't worry about hitting high frequencies can divide up it's logic into fewer stages, simplifying design and requiring fewer transistors for latching.

[1]http://www.computingcompendium.com/p/arm-vs-intel-benchmarks...

The difference is actually smaller, since we're comparing to a 1.3 Ghz Air, not a 3.3 Ghz (or whatever) desktop chip.

The 4250U I mentioned is the one in the 1.3 Ghz Air.

> for Apple's A7 it's around 3x versus a normal desktop chip like the 4770 and only 1.7X versus the 4250U you'd find in a Macbook Air

That would still be noticeably slower for native software, and unbearable for anything run under emulation. It would also be very painful for use cases such as people wanting occasional Windows applications or Linux VMs.

Right, using an A7 in a Macbook would be totally crazy. The idea is that Apple might create an A8 that's about as fast as the current Macbook Air and use that. Getting as much performance increase going from A7 -> A8 as they did going A6 -> A7 strikes me as implausible, and it would probably be a bad idea to switch architectures anyways. But it's not totally crazy.

I think GeekBench is a poor benchmark, since it's measuring peak CPU performance in a narrow domain, but the table you cited shows that the 4770 is 6.5x faster than the A7 (16774 vs 2564). Even at the same frequency and core count, Haswell is 1.8x faster in the benchmark.

Whatever the exact number, it doesn't invalidate the point I was making: Switching to ARM will ruin performance for x86 programs, which need to be emulated. ARM has to be significantly faster than x86, or else the emulation overhead will make users hate the transition.

This is a huge challenge for the Mac Pro and rMBP, which are all about performance. Generally, a given microarchitecture can only scale TDP by a factor of 10 by changing frequency and number of cores. Satisfying the pro line would require a new microarchitecture. Considering an x86 to ARM transition would take 1-2 years, that's asking a lot of Apple's chip designers. If Apple was going to go through all the trouble of building a new ARM core for high-performance, high-TDP applications, they might as well build their own x86 CPU. It would save a transition and the necessary emulation overhead. They could roll it out for the MacBook Airs first, then address the Pro line at their leisure (or not and just use Intel CPUs).

Which is why I think some of where AMD is looking to go is very interesting.

I was looking at the single threaded numbers because those are usually more relevant to what you would be doing on most Macbooks.

Citation needed.

https://en.wikipedia.org/wiki/Amdahl%27s_law perhaps. Single threaded speed is most relevant to software because it's difficult to make it take advantage of multiple cores.

You're making the incorrect assumption that apple would in fact emulate. As they control the whole stack, they can (relatively) quickly switch architectures. They'd lose the market that does in fact install windows on mac hardware, though that is likely a negligible niche. So for the vast majority of their customers there would not be a performance penalty. Quite the opposite.

Apple chips are in terms of tech already competitive with anything Intel has to offer [1], though of course currently optimised for mobile. Introduce a dedicated chip for macOS without the heat and power restrictions and Apple won't be chained to intel any more. It will save the extra $50-500 they pay intel for every chip. On top of that, the battery will be smaller/cheaper; it will require fewer external chips as the SoC will incorporate this functionality (also cheaper, more control); the retina graphics will be driven by a chip that is adequately powered and optimised (intels graphics chips are extremely slow): all of the sudden the whole package is extremely compelling.

All you lose is windows compatibility. A small downside to multiple big upsides.

[1] http://www.anandtech.com/show/7910/apples-cyclone-microarchi...

No. You also lose compatibility with every game and most major applications like Adobe, since all of these use lots of highly optimized, handwritten assembler code for critical sections, as well of all code paths using x86 vector instructions or other hw acceleration.

Thanks but no thanks.

Those vendors will follow. As they have previously when they transitioned from powerpc to x86. Games is a good point. But this is not a big thing on MacOS (as yet) anyways. If it saves hundreds of dollars per machine, its a no-brainer for the vast majority of users.

It took a VERY long time for the large ISVs to switch, and while games aren't big on OS X many people like the fact they can boot into Windows to play them.

If Apple was able to get the performance of their chips up to the level of Intel chips, would they be able to save hundreds per machine? Intel has some very large economies of scale.

If Apple wanted to save hundreds per machine they could do that right now. Intel's marketing campaigns that Apple doesn't take part of (like the 'Intel Inside' stickers) come with large rebates. Apple could use that and not have to go through the huge pain of another architecture switch.

Why should they?

Intel chips blew the doors off PPC by the time Apple got around to the PPC -> Intel transition. That (and good emulation software) allowed a relatively pain-free transition for most consumers: PPC apps didn't feel any slower when run on a new Intel-based Mac. That made it fairly easy to accumulate a critical mass of users with Intel systems, yet it still took ages for native apps to become available from the likes of Adobe and others.

Fast forward to today.

ARM chips are, at best, a little slower than the Intel chips they would replace. On top of that, the x86 ISA has a reputation for being hard to emulate well. So put yourself in the shoes of the consumer: are you going to upgrade to the new ARMBook Air, if it runs all your old apps at 1/2 or 1/4 the speed (and I'm being super optimistic here) they ran on your old MacBook Air? No! You're not! You're going to stick it out until the second generation of ARMBook comes out.

That means no critical mass of people who own ARM-based Macs and are willing and able to pay for native ARM software. That means vendors will. not. follow. This doesn't even take into account that Apple have burned their bridge with Adobe (a major vendor of 3rd party software for Mac).

Last point: I think you're wrong about games. There are tons of games for Mac on Steam. Would vendors go to the trouble of making a Mac port if there was no money in it?

>You also lose compatibility with every game and most major applications like Adobe

Just like they lost compatibility when they switched from PPC to Intel?

I think that's one thing that a lot of commenters in this thread are ignoring: Apple has done this before. They've switched from PPC to Intel when IBM wasn't committing to making the power efficiency gains that they were looking for, and they'll switch from Intel to their own homegrown ARM chips for essentially the same reason.

That switch took years and years and YEARS, and it was done for clear benefit. Desktops were one thing but in laptops Apple was way behind in performance. The new machines were so much faster that the emulated software didn't feel slower in normal use and you gained access to Windows software (which was horrendously slow when emulated on a PPC).

I'm not sure they could do such a transition again on the desktop. Other commenters ideas of making an 'iPad pro' and coming at it from that direction seems like a much better idea if you think any of this should be done.

Don't forget that Apple is up against Intel. They're called Fabzilla for a reason. Apple has a done an amazing job with their phones and tablets but when it comes to raw performance on the desktop I don't think they'd be able to keep up when Intel decided to start playing hardball.

Intel is failing to deliver their existing roadmap on time. Is this because they just haven't 'decided' to play hardball?

They're having issues right now, but they'll pull back. We've been through this before. Look at how far Intel's chips have come from just the Core2 processors of a few years ago.

Either way, this is nothing compared to the mess of delays the G4 and G5 went though, especially on mobile. Compared to what Apple has seen before this is a tiny detour.

And of course since Intel is the market leader, Apple is (at worst) in the same boat as everyone else. With the G4/5 they were falling further and further behind each year.

Well it's certainly plausible that this is the case... They have no meaningful competition at the top end of performance, so clearly they're don't need to bother dumping as much money into R&D as if they did. If they've found that a basic level of incremental improvements maximises profit, then they'll do that. And at the moment they're held up by fabrication issues, sure, but die shrinkage is an important part of their 'tick tock' development model so it makes sense they'd be waiting for this, it's hardly an indication that they're struggling in the grand scheme of things, or that they will fail to deliver the same steady performance improvements moving forward.

I don't think they would snap their fingers and every Mac they sell would abruptly switch to ARM. If they chose to go with ARM they would almost certainly move more gradually and come out with only one or two ARM machines at first. Maybe an iPad "Pro" or something like that which leverages all the software already written for the iPad and iPhone but has a detachable keyboard can also run Mac software if it's recompiled... something like that.

They'd also lose compatibility with any 3rd party hardware (think ThunderBolt), any video editing device drivers which were made for the x86 platform will need to be recompiled for the ARM platform.

A switch like this would kill Thunderbolt and many USB drivers.

Besides, we already saw what happens when Microsoft tried to switch to ARM through WinRT. That was an absolute disaster. Why would Apple try to do the same?

Video editing device drivers might be a problem but I'm pretty sure Apple provides the Thunderbolt drivers so a recompile won't be a problem for that if they make the (very unlikely imo) move to ARM. Correct me if I'm wrong on the Thunderbolt drivers point.

You're making a big assumption. Sure, some apps have handwritten assembly, but that's pretty rare these days. Apple's Accelerate framework (which has been around since the PowerPC days) allows apps that need high performance in certain areas to do that without resorting to assembly:

>This document describes the Accelerate Framework, which contains C APIs for vector and matrix math, digital signal processing, large number handling, and image processing.

https://developer.apple.com/library/mac/documentation/Accele...

What if Apple introduced workstation-class ARM another way; made a really powerful iPad/iPhone which could sit in a dock with keyboard/mouse/monitor, and run both iOS and ARM OSX?

x86 computers could continue to exist for high-end users, but typical-users might be content with a hybrid tablet/computer. As ARM increased in power, x86 might dissapear entirely.

Your comments on Windows compatibility/virtualization are spot on. Cloud streaming, Citrix/XenApp, can help in some situations here. In a few years virtually all major apps will probably be cloud/browser based (Microsoft Office, Adobe Suite, streaming games/apps, etc). Windows on Mac might not be as important then. It's possible Microsoft might release ARM Windows too (besides RT). If ARM gets popular in the datacenter you may see Windows Server 2015 ARM Edition. They've done this in the past with Itanium.

I think that it would be more likely that they'd just drop OS X in that scenario, and just have a beefed-up version of iOS with better keyboard and touchpad support.

The number of users who would actually benefit from OS X vs iOS, but would not be ticked off at an inability to dual boot Windows or run any legacy applications, is very small indeed.

iOS for ARM and OS X for x86 seems likely to remain in the future, but I do think that Apple could do a netbook or dockable tablet running ARM/iOS if they really thought there was demand. I'm not sure there is, but if the product was good enough they have shown in the past an ability to manufacture demand where it didn't exist previously...

What you're describing (a mobile device that docks into a workstation environment and powers the KVM) sounds like the dream mobile product designers have been having for well over a decade. I remember Canonical recently coming up with a concept of a phone that would do this (see: Ubuntu Edge[0]). That being said, the roadmap implied by iOS 8 and Yosemite strongly suggests that they're just going to bridge these gaps over the Internet. Google is also pretty clearly going in this direction, and given the ubiquity of Internet access (especially compared to available KVM terminals), I think this is the direction we're all going to go in.

0: http://en.wikipedia.org/wiki/Ubuntu_Edge

It's been done several, dare I say many times. Badly, of course. I saw one at fry's where the base even had a more powerful CPU and when you docked it, the tablet was just the monitor... but you could access the tablet functionality while docked through something that looked like your typical telivisions "Picture in Picture" functionality. There was also a phone, a while back, that would dock into a video/keyboard in a laptop form factor. (Many phones these days, I am given to understand, have hdmi out, so this is largely a mechanical engineering problem. Well, that and making the software usable on both screens.)

So yeah, if it is, in fact, a useful form factor (and I have my doubts) it's in the perfect place in the technology curve for apple; the tech is all there, but nobody has implemented anything usable.

The discounts aren't that "huge" when compared to ARM chip prices. They are only "huge" in comparison to other Intel customers. But the difference still remains something close to an order of magnitude.

It's been shown before that Intel chips make up roughly 40 percent of the BOM/retail price (however you want to account it) of a PC. I doubt it's less than say 30 percent for an Apple Mac.

The retail price for a high-end chip is like $30, max. The retail price for a Core i5 can be more than $300. So yes, I think the pricing difference can be significant. Or rather the profit difference for Apple.

Apple could basically sell an ARM-based Macbook Air for $700, and make the same amount of profit per unit - BUT, sell a lot more of them, since they're $700 per unit. Apple could also sell it for $800, and make a lot more in profit per unit, too.

Since when Apple cares about their bottom line?

"The retail price for a high-end chip is like $30, max."

Citation?

Microsoft tried to sell WinRT based ARM tablets at $300, and still couldn't sell any. In comparison, the $1000+ Surface Pro (once you get the keyboard attachment and everything, it breaks $1000 easy) has been getting pretty decent reviews and has fallen into a solid niche.

Why would an ARM-based Mac do any better than WinRT?

The transition also worked because they had done it before. NEXTSTEP was already running on several CPU architectures (both big- and little-endian) including x86 before Apple even entered the picture. Undoubtedly Mac OS X is already running on ARM in Apple's labs.

I don't think boot camp/Windows virtualization is relevant. I don't know anyone who uses these features on their Macs. I'm sure there are people for whom this is a make-or-break capability, but it may be that Apple thinks that at this point, they are expendable.

Processor speed means a lot less than it used to. Even processors from half a decade ago are more than fast enough for what most people do with a computer.

I would also venture a guess that Apple may know a percentage if not even a majority are only interested in trying a Mac because the x86 proc gives them a bridge from Windows. If you take that away then mac sales may take a hit and possibly a significant one.

Anecdotally I know for me, this would cause me to move from a top of the line Macbook Pro Retina to an Air or even a mini. Time and market has changed to where having a mac is a must for me as a professional developer but I won't spend 3K on a single machine whose primary differentiator is ability to compile iOS applications.

> It would be worse in almost every aspect.

I don't see any explanation in your comment of how any of those aspects would be worse.

You say "Apple wouldn't profit much more from the switch" because of the volume discounts they get from Intel, but you don't cite exactly how much of a discount they get, and you still concede that Apple's own ARM chips might have a lower cost.

You say "battery life would go up a bit," and seem to imply that it could be up to 20%, which in my opinion is massive.

And you don't address size.

How then are any of those aspects worse? Even under your own seemingly pessimistic estimates, cost and battery life would improve.

Worse from the customer perspective.

All their old apps now run probably 2 to 10 times slower, until new native binaries arrived. Compare that with the PPC->Intel transition, when old binary apps stayed about the same or maybe got a little faster.

Also, emulating Intel binary apps with the CPU pegged for twice as long nullifies any energy savings, so the battery life in practice gets worse. Even if it is 20% for native binary apps, consumers are going to see "new computer, worse battery life, Og smash!" during the transition.

You're not going to see a 20% gain in battery life for native binaries, though. The only way you'd see that is if the ARM CPU used no energy at all. Extraordinary claims, evidence, and all that.

Is there really any reason to think that an Intel and ARM chip of comparable performance will have significantly different package sizes? Intel has always been the leader in process technology, so it would surprise me if the ARM die were any smaller.

You're forgetting about there being no fan.

No fan noise, no fan volume, no openings for dust to get inside the system. I personally can't wait until laptop fans go away.

ARM doesn't mean "no fan". I can't wait for laptop fans to go away but currently the only fanless laptops you can buy are atom based and all of them outperform their ARM equivalents.

When you start getting into "laptop level" power (ie comparable to pentium), even at the ARM level, you start seeing fans. For example, the first generation nvidia shield.