Discussion Apple Silicon SoC thread

[senttoschool]

The point I was making is Apple's cadence depends on its machines, not (just) its chips

We don't expect Apple to introduce a new design every time a new generation of M series comes out. Apple has thus far introduced new machines with new SoCs because Apple Silicon has drastically different thermal requirements than Intel Macs.

Going foward, I expect Apple to introduce brand new designs for each Mac segment once every 4-5 years. Thus, we'll have 4-5 years of pure SoC updates only.

 

[Eug]

We don't expect Apple to introduce a new design every time a new generation of M series comes out. Apple has thus far introduced new machines with new SoCs because Apple Silicon has drastically different thermal requirements than Intel Macs.

Going foward, I expect Apple to introduce brand new designs for each Mac segment once every 4-5 years. Thus, we'll have 4-5 years of pure SoC updates only.

I’m thinking they may update the form factor of the 13” MacBook Pro. They seem reluctant to discontinue that tier.

 

[IntelUser2000]

Whereas at Intel, they have more heated arguments than useful constructive discussions and if they end up making a mistake, Intel Israel is always ready with something to dig them out of their hole.

Uh, no. The last few cores are designed from the Haifa, Israel team.

Also the people responsible for amazing chips in Apple have left. That's big reason why the last few chips have shown less than expected gains.

So what do those people do? They go work at other companies. It's not always at direct competitors, and it could be a startup. But in lots of cases they do switch to direct competitors and it literally makes the others better.

These seasoned veterans of engineers are very limited in numbers so by moving from one company to other you could say one company is "sucking" the lifeforce out of other to become better.

 

[igor_kavinski]

Uh, no. The last few cores are designed from the Haifa, Israel team.

I guess Intel has some sort of non-compete clause for their top Israeli CPU design engineers. Doesn't make sense that they haven't jumped ship to Apple.

 

[Mopetar]

Uh, no. The last few cores are designed from the Haifa, Israel team.

Also the people responsible for amazing chips in Apple have left. That's big reason why the last few chips have shown less than expected gains.

So what do those people do? They go work at other companies. It's not always at direct competitors, and it could be a startup. But in lots of cases they do switch to direct competitors and it literally makes the others better.

These seasoned veterans of engineers are very limited in numbers so by moving from one company to other you could say one company is "sucking" the lifeforce out of other to become better.

Apple is one of the richest companies on the planet. If they really felt that there were a few dozen people who would let them make a chip significantly better than they otherwise can, they can pick up the phone and get those people on the job within weeks.

If that's all it was, Apple would have done it. Pay 100 engineers another $250,000 each every year and you're only spending an extra $25 million. Chump change for them.

 

[IntelUser2000]

If that's all it was, Apple would have done it. Pay 100 engineers another $250,000 each every year and you're only spending an extra $25 million. Chump change for them.

Money isn't top priority for the top engineers. They got better things to aim for like achieving more ambitious designs or just going their own way with the project. Otherwise why do you think they bother with a startup?

So why do you think Apple did better? Just pouring more money at it? You cannot fuel innovation with money. It's ideas that do. It just happens to be good money follows good ideas. It's not just technical talent. It's the ability to inspire, harmonize and focus the group towards a project that results in success. Some organizations take years to find that, some never do.

Why did/is Intel doing so bad especially in sectors like Server? It's certainly not lack of money. "Oh because of braindead management" or "no foresight". But that just supports my point, that it's the people that matter.

Some were saying Apple is starting to have internal issues and pointing out it as the cause for various people within the company leaving.

I guess Intel has some sort of non-compete clause for their top Israeli CPU design engineers. Doesn't make sense that they haven't jumped ship to Apple.

Everyone has their own reasons for staying/leaving. Just saying Haifa being this savior within Intel is not true. The Oregon team was responsible for the integrated memory controller, QPI, Hyperthreading and things like power management unit within Nehalem. Yes they sound "boring" but they play just as an important role.

And Haswell was their project too. Brought the biggest battery life gain in their history.

 

[guidryp]

Money isn't top priority for the top engineers. They got better things to aim for like achieving more ambitious designs or just going their own way with the project. Otherwise why do you think they bother with a startup?

They bother with startups precisely for the money. The chance to make millions in short period of time.

 

[Roland00Address]

Likewise money can mean different things with what time horizon and consistency the person who wants the money, what is their preferred desire.

Many family engineers want a consistent but well off pay check. Others may want to take on more risk for a higher reward. Others may prefer other form of consistency like good pay but they move less and they remain in good schools for their kids.

It is not a one size fits all.

 

[Doug S]

You can tempt good engineers with better pay, but to get the best engineers you may need more. They might want a big challenge like going after a new paradigm or at least a new market, or at least be given more autonomy over the design's strategic direction.

Iterating on the 9th generation of Apple's custom iPhone SoC was probably less attractive than doing the first generation of Apple Silicon. Taking a risk like VLIW/EPIC is going to be more exciting for them than iterating x86 (yeah Itanium flopped, but most big new ideas do)

In some ways a big company with nearly unlimited resources like Apple or Intel might be more limiting - because being so big means failure costs a lot more. Sure they can afford a skunksworks type project off to the side, but that probably will never see the light of day. You have to found a startup if you want to explore your big ideas that don't fit with the plans at a company like Apple or Intel where the end product's needs dictate what you're allowed to design. And if money is a motivation, you can make a lot more money if you're on the ground floor of a startup - even if it fails!

 

[richardskrad]

I don't see why Apple is doomed because they lost great engineers. Apple has in a short amount of time, only a few years, been spreading their employees to focus on:

M1 / M2
M1 Pro / M2 Pro
M1 Max / M2 Max
M1 Ultra / M2 Ultra
M2 Extreme

And also R&D for M3, M4, M5, etc.


These are groundbreaking chips in their respective segments and Apple is doing it all at once. You can have all the best engineers at disposal but as a company, you only have finite bandwidth.

I expect Apple Silicon on 3nm to be a massive leap in performance and efficiency again. The M1/M2 are still the kings of performance per watt. Apple is in no rush to get to 3nm. They are still ahead of Intel and AMD at this moment. I think Apple knows it.

 

[Doug S]

I don't see why Apple is doomed because they lost great engineers. Apple has in a short amount of time, only a few years, been spreading their employees to focus on:

M1 / M2
M1 Pro / M2 Pro
M1 Max / M2 Max
M1 Ultra / M2 Ultra
M2 Extreme

And also R&D for M3, M4, M5, etc.


These are groundbreaking chips in their respective segments and Apple is doing it all at once. You can have all the best engineers at disposal but as a company, you only have finite bandwidth.

I expect Apple Silicon on 3nm to be a massive leap in performance and efficiency again. The M1/M2 are still the kings of performance per watt. Apple is in no rush to get to 3nm. They are still ahead of Intel and AMD at this moment. I think Apple knows it.

I agree, I've been making the point for some time that Apple has been focused on launching ARM Macs, scaling up to the Pro/Max designs, developing that fancy I/O interconnect with tens of thousands for even bigger Macs (the largest of which hasn't launched yet)

So it is premature to suggest the rather minor updates from A14->A15->A16 indicate some big issues with their design team due to GW III et al leaving. They would obviously have the best people focused on launching Apple Silicon, and have reason to be more conservative with core iterations during that time.

Presumably all that AS work has been done for some time now - while the 4 SoC Mac Pro hasn't yet launched the design of the chips would be complete 18-24 months before the Mac Pro launches so they've been able to resume focus on core designs. Perhaps not in time for A17 generation cores, which will have already taped out, but it'll benefit from N3.

 

[nicalandia]

Highest result for the M2 Max

Mac14,6 - Geekbench Browser

Benchmark results for a Mac14,6 with an Apple M2 Max processor.

browser.geekbench.com

 

[richardskrad]

Oh wow, Apple reaching 2000 in Geekbench single-core would be impressive.😯

 

[nicalandia]

Oh wow, Apple reaching 2000 in Geekbench single-core would be impressive.😯

They just did. But So will AMD and Intel top of the line Mobile CPUs

 

[scineram]

With 3 times the wattage.

 

[nicalandia]

With 3 times the wattage.

65 Watts is not that bad

 

[scineram]

A single core? That's horrid!

 

[nicalandia]

A single core? That's horrid!

65 watts per CPU. Zen4 can be very efficient.

 

[nicalandia]

Look at the performance of these guys capped at 65 watts







I know it's Windows vs MacOS but this is the closest we have

 

[Ajay]

Look at the performance of these guys capped at 65 watts

View attachment 72541

View attachment 72543

View attachment 72544

I know it's Windows vs MacOS but this is the closest we have

Hmm, Apple M2 Max isn't out yet. Some dev/partner is having fun (considering Apple's close to the vest approach on these sort of things). Also, any clue what the power usage was for the MacOS run?

 

[nicalandia]

Hmm, Apple M2 Max isn't out yet. Some dev/partner is having fun (considering Apple's close to the vest approach on these sort of things). Also, any clue what the power usage was for the MacOS run?

I would say close to 40 Watts

 

[senttoschool]

Look at the performance of these guys capped at 65 watts

View attachment 72541

View attachment 72543

View attachment 72544

I know it's Windows vs MacOS but this is the closest we have

My M1 Pro runs the Geekbench5 single-core benchmark using 0.3w - 5w. For most of the duration of the test, it uses way less than 5w.

So if AMD/Intel's best mobile chips use 20w-35w to run the same benchmark, then the M series should be about ~7x more efficient.

 

[senttoschool]

I’m thinking they may update the form factor of the 13” MacBook Pro. They seem reluctant to discontinue that tier.

I don't think the 13" Macbook Pro will be here after next year. Apple didn't have anything to slot in between the Air and the 14" Pro so they retained the 13" Pro for far longer than they should. Also, the 13" Pro has a well oiled supply chain which was what Apple needed during covid and supply issues.

Reports are that a 15" Air is coming out next year to go in between the 13" Air and the 14" Pro. That would make much more sense as a lineup and Apple can kill the 13" Pro once and for all.

 

[uzzi38]

My M1 Pro runs the Geekbench5 single-core benchmark using 0.3w - 5w. For most of the duration of the test, it uses way less than 5w.

So if AMD/Intel's best mobile chips use 20w-35w to run the same benchmark, then the M series should be about ~7x more efficient.

Mobile Zen never consumes >20W for pure ST loads. Desktop does, mobile is usually around 15W. It also comes with lower peak performance though, so that's to be expected.

For iso-performance Zen 4 will likely be significantly more competitive with Apple's current end cores in performance/W. A lot of power is used for that last roughly 15% performance. To put things into perspective, Zen 4 uses similar levels of power for 4.5GHz as Zen 3 uses for 3.5GHz. That being said, AMD still have a significant disadvantage on power coming from their uncore by comparison however - what I said only applies to looking at the core power only.

 

[scineram]

65 watts per CPU. Zen4 can be very efficient.

And very means at least 3 times worse than M2.