Live data from Hacker News

Ask HN: Can competitors catch up to Apple Silicon?

news.ycombinator.com

261–270 of 323 posts

Re: Ask HN: Can competitors catch up to Apple Silicon?

#261

You're looking in the wrong place. The magic comes from TSMC, not Apple. Apple's major innovation is in using their unrivaled bank account to pay TSMC for exclusivity on their newest fabrication technology. As competitors gain access to that TSMC technology they will match or beat Apple's performance. And if Intel succeeds in reclaiming the fabrication technology lead from TSMC (we'll see) then Intel will beat Apple'…

This is just absolutely incorrect. Design is over half the battle, look at Qualcomm n5 performance vs Apple M1 on n5 performance. It's not even comparable in anyway. Even AMD 6900U struggles in many regards compared to M1 pro, and even more so at performance per watt. For example Cinebench runs at 118% more power draw on the Ryzen at only a 11% multi-core performance advantage.

> This is just absolutely incorrect. Design is over half the battle

Completely agree. There are many ARM SoCs available in the market, and Apple's ARM SoCs outperform most of them. Apple has also made significant design changes in its SoCs that really does differentiate it from current Intel and AMD chips and these design changes do give it a performance advantage. The unified memory access architecture (I think pioneered by SGI), is a good example of one such architectural difference. And ofcourse, manufacturing these SoCs on newer node tech also allows Apple to pack more transistors in their chip than Intel and AMD currently can. Altogether, it is undeniable that Apple engineers have created a great piece of hardware and a very decent and viable alternative to Intel and AMD chips.

But ... better tech doesn't always necessarily translate to more sales and popular acceptance. AMD is a great example here - despite having better technology than both Intel and Apple (on CPUs and GPUs), it still struggles to compete with either and is nowhere near the market size or capitalisation of either Intel or Apple. However, here we are not talking about a product from a new startup or small business but Apple - A huge behemoth with a fantastic team of management that knows how to sell stuff and make a profit. So it's not at all far-fetched to assume that the Apple Silicon chips are here for the long term and will continue to be a viable competitor to Intel and AMD chips. But how are they competing, and what company's vision will win in the long term?

Apple's long term business vision is to continue being a hardware + software company while evolving & increasing its recurrent revenue from services that is tied to monetising the personal data of its users. I have no clear idea of either Intel's or AMD's future business visions (which is perhaps telling in its own way of how both businesses are today), other than that they want to continuing innovating and selling the hardware that is their core business (CPU, GPU, Fab tech etc.) that is more "open" to integration with other systems (software and hardware).

So, keeping those visions in mind, let's compare the pros and cons of Apple and AMD / Intel chips.

Apple Silicon - Pros:

1. Latest tech, innovative design offering better performance in some areas. 2. Low power usage, great battery life. 3. Cheaper. 4. "Integrated" RAM leading to better performance. 5. Highly integrated with Apple ecosystem (Apple hardware + software + web services).

Apple Silicon - Cons:

1. It is not sold as stand alone part. 2. It is a closed box system with no hardware literature available to build alternative platforms on it. 3. Highly integrated with Apple's ecosystem. 4. Has built-in planned obsolescence (CPU is soldered, RAM is soldered, bootloader can be locked anytime). 5. Low power usage that unnecessarily sacrifices performance is undesirable in a desktop computer.

As is evident, the pros and cons of Apple Silicon are tied to the business vision of Apple - they believe that the best way to make recurrent revenues by selling Apple services, is to create a closed-box system and tightly integrated ecosystem that forces its users to only use Apple hardware and services. Built-in planned obsolescence makes it easier to sell new system. Similarly, AMD / Intel's business vision is reflected in pros and cons of their product too:

AMD / Intel CPUs - Pros:

1. Good support for multiple OS platforms. 2. Backward compatible (x86). 3. More open, better hardware literature on their SoC's support creating new platforms. 4. Better integration with other hardware allows more flexibility to users in building custom hardware. 5. Supports repairability (CPUs can be replaced and upgraded more easily in the same system, integrated GPU doesn't prevent you from adding and using a more powerful GPU etc).

AMD / Intel CPUs - Cons:

1. Lagging in newer tech. 2. Maintaining backward compatibility comes at a price - more complexity, limits exploring new designs and architectures. 3. Supporting multiple platforms can be somewhat limiting as hardware performance varies depending on software quality of the platform.

-----

As AMD and Intel have shown in the past, both have had better tech than the other at some point. It's the same today with Apple innovating and currently having better tech than AMD / Intel. So it isn't exactly a question of will the better tech win. What matters is who can sell their tech better.

Re: Ask HN: Can competitors catch up to Apple Silicon?

#262

You're looking in the wrong place. The magic comes from TSMC, not Apple. Apple's major innovation is in using their unrivaled bank account to pay TSMC for exclusivity on their newest fabrication technology. As competitors gain access to that TSMC technology they will match or beat Apple's performance. And if Intel succeeds in reclaiming the fabrication technology lead from TSMC (we'll see) then Intel will beat Apple'…

what’s your superpower?

I’m really rich.

That justice league exchange is what came to mind when I read your comment

Re: Ask HN: Can competitors catch up to Apple Silicon?

#263
post #241
post #218

Earlier quoted context omitted.

Not really. GPUs regularly implement 256-bit or wider RAM.

Don't GPUs also typically have the RAM soldered on you the card, effectively "on package"?

That's my point; Apple's RAM soldered on package is the same as PC laptops with RAM soldered on the motherboard. Apple has no advantage from that.

Re: Ask HN: Can competitors catch up to Apple Silicon?

#264

Earlier quoted context omitted.

This is just absolutely incorrect. Design is over half the battle, look at Qualcomm n5 performance vs Apple M1 on n5 performance. It's not even comparable in anyway. Even AMD 6900U struggles in many regards compared to M1 pro, and even more so at performance per watt. For example Cinebench runs at 118% more power draw on the Ryzen at only a 11% multi-core performance advantage.

> This is just absolutely incorrect. Design is over half the battle Completely agree. There are many ARM SoCs available in the market, and Apple's ARM SoCs outperform most of them. Apple has also made significant design changes in its SoCs that really does differentiate it from current Intel and AMD chips and these design changes do give it a performance advantage. The unified memory access architecture (I think pion…

> increasing its recurrent revenue from services that is tied to monetising the personal data of its users.

My understanding is that they’re more responsible than other services with respect to personal data. Unless I’m mistaken, their strategy isn't to monetise and sell data, but is instead to sell services that allow users to interact with their own personal data, which is incredibly rich given the ecosystem that Apple Watch and HealthKit offers.

Re: Ask HN: Can competitors catch up to Apple Silicon?

#265

Earlier quoted context omitted.

The other answer was awesome but here’s another one: Number of registers. Of course internally the CPU has some huge number of registers. But it matters how many registers the machine code can speak of, and arm64 has 32 while x86 has 16. Consequently, x86 machine code has more cases of data flow obscured by spills to the stack. In a very sophisticated cpu, that might mean more decoding work to try to reconstruct the…

> X86 throws at least one more limiter just for good measure: the TSO memory model, which causes memory effects to have dependencies between them in cases where they wouldn’t on arm. Apple's M1 supports TSO (on ARM, obviously).

Isn't that only used to support x86 emulation (i.e. apps for the older macs)?

Re: Ask HN: Can competitors catch up to Apple Silicon?

#266

Earlier quoted context omitted.

This is just absolutely incorrect. Design is over half the battle, look at Qualcomm n5 performance vs Apple M1 on n5 performance. It's not even comparable in anyway. Even AMD 6900U struggles in many regards compared to M1 pro, and even more so at performance per watt. For example Cinebench runs at 118% more power draw on the Ryzen at only a 11% multi-core performance advantage.

> This is just absolutely incorrect. Design is over half the battle Completely agree. There are many ARM SoCs available in the market, and Apple's ARM SoCs outperform most of them. Apple has also made significant design changes in its SoCs that really does differentiate it from current Intel and AMD chips and these design changes do give it a performance advantage. The unified memory access architecture (I think pion…

>” 5. Low power usage that unnecessarily sacrifices performance is undesirable in a desktop computer.”

This is a temporary matter. For notebook computers, Apple made the obvious decision to optimize for power usage. But in interviews their engineers have signaled that when the time comes they will be just as capable of optimizing for performance on a future Mac Pro.

Re: Ask HN: Can competitors catch up to Apple Silicon?

#267

Earlier quoted context omitted.

> This is just absolutely incorrect. Design is over half the battle Completely agree. There are many ARM SoCs available in the market, and Apple's ARM SoCs outperform most of them. Apple has also made significant design changes in its SoCs that really does differentiate it from current Intel and AMD chips and these design changes do give it a performance advantage. The unified memory access architecture (I think pion…

> increasing its recurrent revenue from services that is tied to monetising the personal data of its users. My understanding is that they’re more responsible than other services with respect to personal data. Unless I’m mistaken, their strategy isn't to monetise and sell data, but is instead to sell services that allow users to interact with their own personal data , which is incredibly rich given the ecosystem that…

As an Apple user, I see it more cynically now - their whole idea of "privacy" has morphed into a strategy to prevent their competitors access to the personal data of the users, while ensuring they have more invasive access to it. This idea is offensive to me. Preventing data collection is the first step to user privacy. Once BigTech has your data, they can do anything with it in the future (despite what they say today) because of the lack of appropriate law and regulation to prevent this. I see us heading to a dystopian future where BigTech evolve into data brokers.

Re: Ask HN: Can competitors catch up to Apple Silicon?

#268

Earlier quoted context omitted.

That would be a dumb way to build a decoder. Which is why that's not how it works on x86 processors. What an x86 processor does is fetch an entire cacheline of instructions at once. Working out which byte corresponds to instructions is a straightforward state machine--and state machines can be parallelized as reduction tree, because function application is associative! (This is usually described as "x86 processors st…

> fetch an entire cacheline of instructions at once And how many instructions fit into it? Are you going to feed an 8-wide decoder with a single 64B cacheline? Especially with no promises that any instruction starts or ends on that boundary? > It's actually hard to come up with x86 instructions that are 15 bytes long. Something as simple and trivial as loading a 64-bit immediate into a register is a 10-byte instructi…

> Something as simple and trivial as loading a 64-bit immediate into a register is a 10-byte instruction in x86...

It will only take a 10 byte instruction if you can't zero- or sign-extend from a 32-bit immediate. Of course, if you were on ARM or most RISC processors, such an immediate likely requires at least two instructions anyways.

From paging through an objdump output of a large program, I'd hazard that average instruction size is about 4 bytes, maybe a touch shy of that. After all, instructions like "add rax, rbx" is going to be 3 bytes (two if you're only using 32-bit registers!).

> Their current top of the line decoder is IIRC 5-wide.

Looking at wikichip for the block diagram [1], that's 5 instructions per 16-byte window.

[1] https://en.wikichip.org/wiki/intel/microarchitectures/skylak...

Re: Ask HN: Can competitors catch up to Apple Silicon?

#269

Earlier quoted context omitted.

The ASML EUV machines are extremely complicated and rate limited by the polishing of the Zeiss mirrors which require 9 months to manufacture. Zeiss is in the process of scaling up, but in the meantime the EUV machines are very much backordered and that supply limitation is partially responsible for the slowdown in the release of updated nodes. There is an update of the ASML machines which are far more complicated tha…

To continue the tree of suppliers. Zeiss SMT would be nothing without: HAP GmbH Dresden, MBFZ Toolcraft GmbH, Corning Advanced Optics, AGC Inc, RUAG Space, Schmolz + Bickenbach, RORZE CORPORATION and GEWO Feinmechanik GmbH

And those depend on dozens of more companies, and those on dozen more. That is until suddenly a one man specialist retires his company and the entire chains grinds to a halt for months.

Re: Ask HN: Can competitors catch up to Apple Silicon?

#270

You're looking in the wrong place. The magic comes from TSMC, not Apple. Apple's major innovation is in using their unrivaled bank account to pay TSMC for exclusivity on their newest fabrication technology. As competitors gain access to that TSMC technology they will match or beat Apple's performance. And if Intel succeeds in reclaiming the fabrication technology lead from TSMC (we'll see) then Intel will beat Apple'…

Wrong, the majority of Apple’s thermal efficiency comes from superior design which in turn comes from the fact that they were able to design it from the ground up with no backwards compatibility requirements (since they don’t have any customers of the chip besides themselves).

This is false. Apples arm chips are compatible with the normal arm instruction set. Evidenced by Linux trivially running on it.
Post reply on HN