That's not "likely" at all. Apple's CPUs are already faster in single-core benchmarks than the Intel competition, even though the Apple processors in question are designed for mobile devices and have much less power to work with.

Apple would not be making this switch if they didn't think they could improve performance.

That's not remotely true. There is some crossover, which is notable. But that means that the fastest iPhone is faster than some of the slower desktops. Intel's current single core leader is a coffee lake running at 4.7 GHz with like 12MB of core-speed cache. Be real.

Gotta remember that all these previous ARM chips have been designed in the context of the phone. If you suddenly tell the chip designers they have 10x the power and thermal headroom, theres a lot you could do.

Sure, but "could do" isn't remotely the same thing as "already faster" now, is it?

Apple's A CPUs are already the best in their category, by far.

Knowing Apple had Intel Macs running in the background for years before they announced the jump to x86, I wouldn't be surprised if they already have the chips and know how well it performs.

And I also wouldn't mind if they did another demo like they used to: https://youtu.be/oxwmF0OJ0vg?t=24m45s

The current A11 chip is faster single-core than the current MacBook Pro.

The Intel chip you mention isn't in any Apple machine at all. (Except maybe the iMac Pro? Is it even in that?) So you be real. I don't think you've actually looked at the benchmarks. You should do so. I think you'll be shocked at just how much overlap there is, just like I was.

>I don't think you've actually looked at the benchmarks.

What benchmarks? Geekbench? That's the only one I've ever seen where the fight is close, and Geekbench (like everything that boils performance down to one number) is nearly completely useless.

If you see a two-watt phone CPU beating a 45-watt actively-cooled laptop CPU, you can either conclude that the phone is alien technology decades ahead of anything else on the market or that the benchmark is broken. Which is more likely?

https://www.realworldtech.com/forum/?threadid=136526&curpost...

Does Linus not realize that Intel processors also have hardware assisted SHA?

Geekbench may not be the best tool for determining the power of a processor, but at the least it gives you a general idea of power across devices in a consistent set of tests, which can be extrapolated.

They didn't at the time he wrote that.

That post is pretty outdated. It's about old hardware and an old version of Geekbench. But people still use Geekbench 3 and old x86 hardware, so it's not totally irrelevant.

We are talking about the MacBook, IE passively cooled Intel Core m3 running at 1.2 GHz on something like 8 Watts TDP.

that's because apple is fixed on choosing the intel parts with the least tdp power they can get away with. it's their choice to make the mbp slow, up and including the numerous thermal issues that force the cpu out of the nominal turbo mode for the longer workloads

Those are valid points, but it's still shocking that Apple's own CPUs can outperform Intel laptop CPUs with far higher TDP.

Benchmarks aside, because they can and are gamed, the only real way to see this, is loading a big workload such as Photoshop, Kernel compilation, ect., that professionals do every single day. I very much doubt that A11 is even remotely close to a mid range quadcore Intel cpu, and let's not forget that we're in the middle of moving to 6-8 core cpus.

The 8700k which the previous poster mentioned is a hex core with 12 Threads!

> Apple's CPUs are already faster in single-core benchmarks than the Intel competition

Extraordinary claims require extraordinary evidence. "But Geekbench..." is not extraordinary evidence. Give me a board where I can do an Apple to apples comparison, booting both under the same revision of the same OS, and running industry standard benchmarks (read: plural), and then we can come to that conclusion.

No one is stopping you from doing that experiment.

> Extraordinary claims require extraordinary evidence.

"Extraordinary" is in the eye of the beholder. We have some evidence; you're welcome to believe it or not, but unless you supply alternative evidence you're not really making an argument. You're just saying "it might not be true," which is something anyone can say at any time about anything.

Nice, that was a good explanation for your stance. I am not supporting either, just liked the turnout. Obviously this isn't my usual kind of hn comment.

"booting both under the same revision of the same OS"

It should be irrelevant in Apple's case anyway since they run their own OS. If their OS is faster for a certain task than some other OS then that's part of what you would define as "faster" for a user of the device.

Right? Am I missing something?

Faster, probably, but 2x+ Faster like in PowerPC > Intel very unlikely. Which is why there’s much less likely to be enough performance budget for effective emulation.

Presumably their "desktop-grade" A-whatever chip will have a higher TDP and clock speed than their mobile counterparts.

If they could just turn up the TDP and clock speed on their chips and get 2x the performance of Intel's best chips that easily- they would already own the entire desktop market.

And some of us, with actual CPU design backgrounds, have been saying that for a while now.

We've been asking why Apple doesn't already do it.

As someone without a CPU design backround, would something like this scale pretty linearly with added power and thermal headroom? I assume there are limit that would have to be overcome, but what would an ARM chip in the conditions of an i7 look like?

TDP typically scales as somewhere between the cube and fourth power of the clock speed if you're pushing the envelope (in the sense of running at frequencies where further frequency increase also needs a voltage increase). So having 10x the thermal envelope means you can probably clock about twice as fast, all else being equal.

It's more of a logarithmic scale.

The same architecture can generally scale to 10x over a few process generations.

This does not hold at 10nm and below.

Seems that it’s more of a logistical and network-effect issue in getting everyone to support it, rather than a technical issue.

Possibly a patent issue also

My only question is if they actually use ARM for their next-generation architecture, or something completely new...

Computer architectures routinely see 3x performance jumps across different power budgets. This rule has held over decades.

Clock speeds alone can probably increase by 30%. Caches and internal datapaths can double or more. Then you can start to add in more execution units or more expensive branch-prediciton or even new more power-hungry instructions.

A 4 Watt Intel Pentium 4410Y Kaby Lake for mobile devices gets about 1800 on Geekbench, while a 115 Watt Intel Core i7-7700K Kaby Lake for desktops gets 5600.

I'm just going to say it: the Apple laptop CPU is going to get Geekbench score... above 9000!

And, yes, I do have a CPU design background.

So artificial benchmarks already do a very poor job of capturing performance. The apple laptop cpu does not exist. If it did exist it would likely suffer a very substantial performance hit if forced to emulate x86 software. So you are going to speculate on the meaningless benchmark numbers of an imaginary cpu that will take a wholly unknown hit if everyone does not rewrite everything why?

Artificial benchmarks do a great job of capturing performance, since they're more controlled and eliminate unnecessary variables.

Once you understand this, then you can understand how CPU designers work to predict future performance. CPU designers use artificial testbenches.

You making up numbers doesn't appear to be a useful endeavor.

I suspect if Apple designs a desktop CPU, performant x86 emulation will be a key design criteria. I know very little about CPU design, but I imagine it would be possible to have hardware optimisations for x86 emulation just like we have today for video codecs.

Or even further they could bake a "rosetta" into the chip's microcode and have their CPU natively support the x86 instruction set along with ARM or whatever they come up with.

Which is the previous gen and was sandbagged as the 50% increase in cores for coffee lake show.

If someone disagrees they should state why

Do you think they will use the same chip? In 2020?

The processors wouldn't necessarily be less powerful. Case in point is the A11 outpacing the CoreM processors.

The greater threat is the ios-ofcation of macOS. We've seen many companies pull their apps from the Mac App store due to the limitations of the sandbox.

ARM less powerful? Wait and see what Apple can do with their own ARM-design if they target Desktops / Laptops. We might get to see 32 or even 64 cores in a regular MBP. Now tell me how THAT would be less powerful than the current Intel offering.

Apple has always gone down the path of fewer, more powerful cores instead of packing many, weaker ones. Apple's chips were dual core until A10.

And it has let them eke out performance that absolutely leaves almost all Android phones in the dust, and all that with a battery that's 30-50% smaller than the rest (although that does bring its own problems, as we now know.)

Imagine if they're not held back by the battery at all, or even just have to match the power consumption of the power-hungry x86 chips.

Let's be realistic. 64 A72 or A11 cores is a pipe dream. It's not going to happen without a massive nm process advantage.

If we had a MBP with 64 wimpy a53 cores it would still only have 50% of the performance of a Intel i7 8700k or AMD Ryzen 1800X on paper. Of course that is faster than most notebooks but if you truly care about CPU performance then you usually wouldn't use a notebook in the first place.

> In this instance, Apple would likely be going to a less powerful processors…

Arguably, the A11 Bionic is already faster than mainstream desktop CPUs.

"The iPhone 8 even edged out the score from the 13-inch Apple MacBook Pro with a 7th-generation Core i5 processor. That notebook notched 9,213. Is Geekbench 4 really comparable from phone to desktop? According to the founder of Geekbench, John Poole, 'the short is answer is yes that the scores are comparable across platforms, so if an iPhone 8 scores higher than an i5, then the iPhone 8 is faster than the i5.'"[1]

[1] https://www.tomsguide.com/us/iphone-8-benchmarks-fastest-pho...

A 6-core CPU beat a 2-core one in a perfectly scaling multithreaded benchmark? That's just shocking! Truly, that is definitely, unquestionably, beyond any doubt representative of what a battle at the workstation level would be.

But more seriously and less snarky there's a reason they used the 13" macbook pro there specifically and that's because the 15" with the quad-core i7 destroys the A11, 15k vs. 10k. Of course that's with wildly different power & thermal budgets, and I haven't seen much reason to think these geekbench results are at all representative of anything, so grains of salt and all that.

Don’t forget that the current gen A processors are designed to run in something with a tiny battery compared to a laptop. It’s likely that any MacBook bound Apple CPU won’t have such a limitation and will be able to run much faster as a result.

Also if you look at the performance trajectory of A-series in the last five years, 2020 sounds about right for where the crossover will come even between iOS based A-series processors and the fastest of what Intel has to offer, provided they can keep up the pace of performance improvement.

> It’s likely that any MacBook bound Apple CPU won’t have such a limitation and will be able to run much faster as a result.

Geekbench is based on the unthrottled peak performance of the CPU. Increasing the TDP headroom eliminates throttling but Geekbench doesn't measure throttling.

You can perhaps increase performance by overclocking but this requires the CPU to be designed for higher frequencies. AMD's Ryzen CPUs usually top out at around 4 GHz. Intel's CPU can be overclocked up to 5 GHz. The maximum clock speed is limited by the slowest component of the CPU. If the slowest operation takes 0.25 nanoseconds to complete this limits your frequency to 4GHz. If Apple had enough foresight to design their chips with this in mind then maybe but in reality they probably optimized the chip entirely for mobile TDPs.

but the question would be:

will apple up the battery requirements for their own cpu/gpu, or would they simply drop battery while striving for a thinner laptop?

this from someone who's been buying/using Macs for a long time - I'd love to see some very fast arm competition, but is that what it'd really turn out to be?

Why would Intel stand still while arm improves dramatically?

Apple has been gaining on Intel while both have been improving. He’s projecting where those lines cross.

> Apple has been gaining on Intel while both have been improving.

Have they really, though? What lines are actually crossing? Intel's mobile CPUs have continued to shed power, so how much of this is actually Apple "catching up" to Intel vs. Intel just optimizing for power at the cost of performance? As in, is there any performance gap that's actually shrinking, or are phone SoCs just getting more power hungry while laptop ones are getting less power hungry?

Because phone SoCs have gotten rather monsterously power hungry compared to years past. They are actually 5w under real world load parts now, with devices just letting them thermal throttle rapidly to achieve higher burst rates. Laptops, by contrast, are ~10w TDP, vastly less than they were a decade ago.

This would probably make a fascinating in-depth analysis, but the singular data point of a 5w SoC being within spitting distance of a 10w SoC is hardly a revolutionary story. It's pretty much what you expect.

Faster != Faster. Adobe and Autodesk likely have many architecture specific optimizations, and take advantage of the large x86-64 instruction set.

Geekbench is a poor tool. Maybe for simple mathematical operations, but x86-64 has a larger instruction set than what the A11 supports

I dont think there is strong connection between a larger instruction set, and performance.

Using a bad example, let's say x86-64 and A11 can both add numbers up to 3 digits. The a11 can do that faster than x86-64. But I want to add a 5 digit number. The x86-64 can do that with one instruction, but the A11 has to break that 5 digit number into multiple 3 digit numbers in order to process it.

From my understanding, which might be wrong, Geekbench only tests what both can do. Comparing a pickup truck and a sports car you need to look at more than just how fast they can do. You can both move furniture, but the sports car will need to do a lot more to be able to what the pickup can.

On the other hand, Apple hasn't been shy about putting fixed function hardware on the Ax chips to eke out the maximum performance for a given power budget. They were much faster at adding h265 decode/encode than Intel, for instance. I would imagine they'd spend their (larger) transistor budget on a lot of this type of dark silicon that would assist in speeding up common operations.

Geekbench doesn't benchmark single instructions.

This doesn't make any sense.

Yes, Geekbench is a terrible CPU benchmark, but both of them are Turing complete, so there is no functionality difference. And complex instructions are broken into uOps in x86 too.

You are 100% correct.

Faster is not necessarily more powerful.

...in one benchmark, from a heavily Apple-favoring company.

I'm sorry, we need better evidence.

Microsoft is running 32-bit x86 applications on ARM today:

https://channel9.msdn.com/Events/Build/2017/P4171

There are also fat binaries - a recompile for apps that don't do anything too esoteric and you've got both formats in one bundle.

There's no reason to think that the ARM chips they use won't be as powerful as their intel equivalents in a few years.

This is true. The rate of progress in CPU performance has slowed to a crawl and will only get slower in the future. Because the pace of innovation of fabs has slowed Intel has also lost their fab lead and GF/Samsung/TSMC should all be very competitive with Intel over the next decade.

Apple is probably looking at all these trends and concluding that they don't really need Intel.

I'm envisioning a Mac with multiple Ann chips in it. Why stop at 1 when you could put 8 or 14 Ann chips in it?

After the Meltdown patch and Intel's exposed security flaws, Apple might still be better off with less powerful processors.

When Apple switches, their processors are likely going to be 2x faster than Intels. It will be a huge jump.