Discussion Apple Silicon SoC thread

[Eug]

M1
5 nm
Unified memory architecture - LP-DDR4
16 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 12 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache
(Apple claims the 4 high-effiency cores alone perform like a dual-core Intel MacBook Air)

8-core iGPU (but there is a 7-core variant, likely with one inactive core)
128 execution units
Up to 24576 concurrent threads
2.6 Teraflops
82 Gigatexels/s
41 gigapixels/s

16-core neural engine
Secure Enclave
USB 4

Products:
$999 ($899 edu) 13" MacBook Air (fanless) - 18 hour video playback battery life
$699 Mac mini (with fan)
$1299 ($1199 edu) 13" MacBook Pro (with fan) - 20 hour video playback battery life

Memory options 8 GB and 16 GB. No 32 GB option (unless you go Intel).

It should be noted that the M1 chip in these three Macs is the same (aside from GPU core number). Basically, Apple is taking the same approach which these chips as they do the iPhones and iPads. Just one SKU (excluding the X variants), which is the same across all iDevices (aside from maybe slight clock speed differences occasionally).

EDIT:



M1 Pro 8-core CPU (6+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 16-core GPU
M1 Max 10-core CPU (8+2), 24-core GPU
M1 Max 10-core CPU (8+2), 32-core GPU

M1 Pro and M1 Max discussion here:

Page 78 - Discussion - Apple Silicon SoC thread

Page 78 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M1 Ultra discussion here:

Page 109 - Discussion - Apple Silicon SoC thread

Page 109 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M2 discussion here:

Page 127 - Discussion - Apple Silicon SoC thread

Page 127 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

Second Generation 5 nm
Unified memory architecture - LPDDR5, up to 24 GB and 100 GB/s
20 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 16 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache

10-core iGPU (but there is an 8-core variant)
3.6 Teraflops

16-core neural engine
Secure Enclave
USB 4

Hardware acceleration for 8K h.264, h.264, ProRes

M3 Family discussion here:

Page 215 - Discussion - Apple Silicon SoC thread

Page 215 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M4 Family discussion here:

Page 263 - Discussion - Apple Silicon SoC thread

Page 263 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[igor_kavinski]

I think Notebookcheck tested power consumption only on the 10-core variant. I would have liked to see the power consumption numbers on the 9-core variant, since that would make for an interesting comparison, and that is the one I will buy.

If both those models have to meet the same TBP (guessing here), wouldn't the 10-core model be a better-binned SoC with less leaky die that can give the same ST performance at lower power so you get slightly better battery life?

 

[Nothingness]

https://twitter.com/x/status/1787959708156129512

That tweet from an AMD apologist is funny. x86 chips were far behind Apple in IPC a few years ago; Intel stagnated for years; AMD was so far behind that increasing IPC was not difficult. Intel and AMD are only just catching up. Let's see how their next generations fare.

Which doesn't mean Apple doesn't have a problem. But saying AMD or Intel was the right fit for them is ridiculous.

 

[reggie_fils_aime]

Which is going to be a very interesting squeeze. They can't get any more IPC since all the IPC makers apparently left for Nuvia.
We're going to see a very wild switcheroo where QC will provide M1-like chips, for cheaper, and go to M2/3/4 pretty quickly without raising area or frequency too much.
While Apple will have to eat through their margins for area or admit that fans are necessary again because freq has gone up too much.

https://twitter.com/x/status/1787959708156129512

Mmmmmh.

That really is a hot take. The M4 is a chip with one fanless, mobile deployment and it's about as fast in passmark benches as an i9-10900X - and won't be seen outside of machines that are overwhelmingly used as fancy chromebooks. It's not a world-beater, or even terribly novel, but it's... fine? It's new? It still fits inside of everything and barely needs active cooling?

Anyone who wants/needs a Mac is either uninterested in how fast it is vs. Intel or AMD, or knows that there are faster things out there and flat-out doesn't care because MacOS runs well on actual chromebook-class CPUs. A prospective M4 Max/Ultra would be faster than all-but the most expensive x86 processors, which are only available in systems that dwarf every Mac's MSRP (save for the Pro).

The reality of the situation is, I think, is that none of this matters to anyone that isn't us. 90% of people who purchase these things will say "wow! that's a lot faster and cooler than the last Intel one I bought, I'm happy!" and Apple will continue to print money at light speed. That's because the chips do what they need to do, and Apple will continue to use them until they cease to fill that purpose. This is what happened with the 6502, 68k, PPC, and x86 in the Apple world... it may well happen to ARM as well, but at that point we could see Apple go full RISC-V and take full control over their cores.

 

[Eug]

If both those models have to meet the same TBP (guessing here), wouldn't the 10-core model be a better-binned SoC with less leaky die that can give the same ST performance at lower power so you get slightly better battery life?

Probably not. First of all I betcha most are binned for marketing reasons. Second, the deactivated core is a performance core. Third, the power measurements include multi-core.

 

[igor_kavinski]

But saying AMD or Intel was the right fit for them is ridiculous.

What he means is that solely depending on ARM is kind of digging a hole for themselves. What if ARM innovation hits a wall? Then they will have to expend extra resources to switch to some other architecture later on. Maintaining two ISAs with x86 models priced lower could've been one way to prevent future issues with ARM IPC scaling.

 

[reggie_fils_aime]

What he means is that solely depending on ARM is kind of digging a hole for themselves. What if ARM innovation hits a wall? Then they will have to expend extra resources to switch to some other architecture later on. Maintaining two ISAs with x86 models priced lower could've been one way to prevent future issues with ARM IPC scaling.

To Apple's credit, they aren't shy about finding a new place to settle down on when that happens, based on their history.

 

[igor_kavinski]

First of all I betcha most are binned for marketing reasons.

If they are disabling cores just to hit a lower price point, that's mad wasteful. But I guess Apple can do anything. Never hurt them before.

 

[FlameTail]

What he means is that solely depending on ARM is kind of digging a hole for themselves. What if ARM innovation hits a wall? Then they will have to expend extra resources to switch to some other architecture later on. Maintaining two ISAs with x86 models priced lower could've been one way to prevent future issues with ARM IPC scaling.

ISA has nothing to do with IPC.

Have you guys learnt nothing from Jim Keller?

 

[FlameTail]

You guys should see the meme I posted in the Qualcomm thread. (It references Apple).

 

[igor_kavinski]

To Apple's credit, they aren't shy about finding a new place to settle down on when that happens, based on their history.

Their engineers must hate them.

Management: Teams! We are switching to RISC-V now!

Engineers: *GROAN* and maybe even *EXPLETIVES*

 

[reggie_fils_aime]

Their engineers must hate them.

Management: Teams! We are switching to RISC-V now!

Engineers: *GROAN* and maybe even *EXPLETIVES*

One guy in the marketing department: "finally, my RISC-V business/Risky Business pun is coming to fruition!"

edit: to be concise, I'm visualizing the dancing-in-the-socks scene with Yuppies holding Macbooks

 

[igor_kavinski]

ISA has nothing to do with IPC.

It's not about ISA per se. But the way ARM arch has been designed, it can't sustain crazy high clocks like x86-64. It's too thermally limited to allow that.

 

[FlameTail]

It's not about ISA per se. But the way ARM arch has been designed, it can't sustain crazy high clocks like x86-64. It's too thermally limited to allow that.

Yes, it has nothing to with the ISA but everything to do with the microarchitecture.

Because ARM has been used in phones for decades, ARM core designers have followed a design philosophy of "less frequency + high IPC".

 

[FlameTail]

Their engineers must hate them.

Management: Teams! We are switching to RISC-V now!

Engineers: *GROAN* and maybe even *EXPLETIVES*

Did you forget the Devolopers?

 

[Eug]

If they are disabling cores just to hit a lower price point, that's mad wasteful. But I guess Apple can do anything. Never hurt them before.

You mean like everyone else? It’s not as if this is new to the industry. x86 has been doing this for literally decades.

 

[Mopetar]

Switching to RISC-V would be more of a pain for the software side of the company than for the hardware team. The execution engine for their CPU cores probably wouldn't change much at all. Meanwhile they've got to tweak all of their software, drivers, etc. and get their compilers to do as good of a job at spitting out RISC-V as it does ARM instructions. The two do have a lot of similarities though, so I don't think it would be as big of a move when they got off of x86 on their Macs.

I don't know how much money Apple spends on licensing the ARM ISA, but if it ever did become large enough I could see them making yet another architectural transition.

 

[Mahboi]

That tweet from an AMD apologist is funny. x86 chips were far behind Apple in IPC a few years ago; Intel stagnated for years; AMD was so far behind that increasing IPC was not difficult. Intel and AMD are only just catching up. Let's see how their next generations fare.

Which doesn't mean Apple doesn't have a problem. But saying AMD or Intel was the right fit for them is ridiculous.

That's not what he said at all LOL.
He said that Apple moved to their own designs (which ofc would be ARM based) and that in hindsight, their management/abandonment of the design team would lose them everything.
They now have cores with a huge area and can't make good IPC increases since they've lost their teams. Competing is going to be next to impossible right now, because they'll be having a large stake in being ARM-based, while having no options but to increase area even further or raise freq.
Just going back to Intel/AMD is going to be bad, and they can't improve their IPC at all, haven't for years. I'm going to laugh so hard if in a few years they're forced to buy QC, that would be the ultimate irony.

 

[igor_kavinski]

I'm going to laugh so hard if in a few years they're forced to buy QC, that would be the ultimate irony.

Then they manage exactly one ground-breaking architecture in about 4 or 5 years and the core team exits all together to a new startup!

Clearly, Apple management is not paying their engineers what they are worth in terms of how much money they are making for the whole company. Classic dumbass management vs. the real heroes situation.

 

[igor_kavinski]

You mean like everyone else? It’s not as if this is new to the industry. x86 has been doing this for literally decades.

For cheap chips, yes. Apple chips are not exactly cheap to make and certainly don't end up in cheap devices, unlike x86.

It seems more criminal in Apple's case.

 

[Mahboi]

That really is a hot take. The M4 is a chip with one fanless, mobile deployment and it's about as fast in passmark benches as an i9-10900X - and won't be seen outside of machines that are overwhelmingly used as fancy chromebooks. It's not a world-beater, or even terribly novel, but it's... fine? It's new? It still fits inside of everything and barely needs active cooling?

Anyone who wants/needs a Mac is either uninterested in how fast it is vs. Intel or AMD, or knows that there are faster things out there and flat-out doesn't care because MacOS runs well on actual chromebook-class CPUs. A prospective M4 Max/Ultra would be faster than all-but the most expensive x86 processors, which are only available in systems that dwarf every Mac's MSRP (save for the Pro).

The reality of the situation is, I think, is that none of this matters to anyone that isn't us. 90% of people who purchase these things will say "wow! that's a lot faster and cooler than the last Intel one I bought, I'm happy!" and Apple will continue to print money at light speed. That's because the chips do what they need to do, and Apple will continue to use them until they cease to fill that purpose. This is what happened with the 6502, 68k, PPC, and x86 in the Apple world... it may well happen to ARM as well, but at that point we could see Apple go full RISC-V and take full control over their cores.

If we were in 2022/2023, I'd agree. But the problem is, 2024 is the year of (the paper launch of) Lunar Lake. 2025 will be the year of Z5 LP.
The point isn't that the perf is better or worse. The point is that the companies that can do more in terms of perf, can do less for less watts. Apple can't do what Intel/AMD can, but Intel/AMD can do it, and if the latest Skymont rumours prove true, they already effectively have. Without news on Z5 LP yet, we don't know how well will AMD do, but they haven't disappointed with CPUs in a looooong time.

If Apple sells a large area, highly efficient chip, and it works for their business, within a year or so, Intel/AMD and QC will all have M1/M2 tier stuff. If they wanted to make a really phat SoC with lower frequencies to get more perf but not too much wattage, they could get that done very well too.
Apple is now in their little corner where they have no personnel to raise their IPC, and have proven that the low power decent perf market is real. They showed everyone how big that market was, and lost the ability to compete further with the others.
Not good.

 

[FlameTail]

I am one of those who believe that Apple will eventually get their CPU team in order, and release a brand new P-core architecture with a huge IPC improvement. I thought it would happen with M4, but it didn't. So I am holding out for M5.

 

[Nothingness]

That's not what he said at all LOL.
He said that Apple moved to their own designs (which ofc would be ARM based) and that in hindsight, their management/abandonment of the design team would lose them everything.
They now have cores with a huge area and can't make good IPC increases since they've lost their teams. Competing is going to be next to impossible right now, because they'll be having a large stake in being ARM-based, while having no options but to increase area even further or raise freq.
Just going back to Intel/AMD is going to be bad, and they can't improve their IPC at all, haven't for years. I'm going to laugh so hard if in a few years they're forced to buy QC, that would be the ultimate irony.

We certainly don't interpret his tweet the same way.

I stick to my point: back when they switched they made the right decision. Claims in hindsight are pointless. And even now we can't be sure AMD and Intel will overtake them in efficiency. We only have rumors at this point.

And as I wrote in the part you omitted in your quote: yes Apple seem to have a problem at the moment. But even in hindsight that doesn't make their decision wrong. What did Intel and AMD have to propose back then that would have matched what they already had in-house?

 

[Nothingness]

Clearly, Apple management is not paying their engineers what they are worth in terms of how much money they are making for the whole company. Classic dumbass management vs. the real heroes situation.

I don't think that's the problem: Apple engineers are very well paid. The issue is that talented engineers want to have fun on top of money. And the Apple team wanted to go to the server space which Apple didn't want. So they left and funded Nuvia. Only to be bought by Qualcomm and return to laptop chips

EDIT: And I'm glad Nuvia does laptop chips

 

[Mahboi]

I stick to my point: back when they switched they made the right decision. Claims in hindsight are pointless.

That's fair, but the fault was to underpay/not keep their design teams then. Same result, different analysis. And I agree with Kepler's, in the end, they've pushed themselves into a corner.

And even now we can't be sure AMD and Intel will overtake them in efficiency. We only have rumors at this point.

I wanna respond with this:

AMD Ryzen 5 5600X vs AMD Ryzen 5 5600U vs AMD Ryzen 5 5600 vs AMD Ryzen 5 5600H [cpubenchmark.net] by PassMark Software

The problem is obviously that (even counting that TDPs aren't fully respected), there is a large discrepancy between how AMD/Intel have been playing with their V/F curve and how Apple has.
5600X is the "65W" standard for AMD.
But if you pick the exact same CPU and lower it to ~15W, you lose only 30% in MT and 13% in ST. This is 1/5th the wattage for 87% performance in ST, which is the cornerstone of a broadly pleasant experience. You don't want to wait for webpages to get rendered or programs to load too slow.

Apple has been aiming from the start at something like a 5600u. AMD and Intel didn't see the lower power/no fan market as that interesting to them. But it was never particularly hard to do, especially for AMD. People think of Apple as "capable of strong CPUs with no fans", sorry, but there is no way that's not the case for the other two or QC, if you can do more you can do less. It was just a question of showing that it was a worthwhile market, and it so happens that again, both Lunar Lake and Z5LP will be out within a year or so.

Also, I picked Z3 over Z4 because Z3 was a rework of the core while Z4 was more of a tweak and shrink. Z5 is also a rework, and is certain to widen quite a lot. It's going to be even more low frequency capable than Z3 was.

And as I wrote in the part you omitted in your quote: yes Apple seem to have a problem at the moment. But even in hindsight that doesn't make their decision wrong. What did Intel and AMD have to propose back then that would have matched what they already had in-house?

Four years is way more than a moment. Sorry but that's a heavy cope.
And again if you want to say "they were right at the time", sure, but don't forget that it would've been just as possible to seek a Z2/Z3 lower power SoC. They chose to go ARM and do it themselves, then didn't manage to retain their design teams. That will eat them from the inside.

 

[roger_k]

He said that Apple moved to their own designs (which ofc would be ARM based) and that in hindsight, their management/abandonment of the design team would lose them everything.
They now have cores with a huge area and can't make good IPC increases since they've lost their teams.

Or maybe it's just difficult to make good IPC increases if one already leads the market in IPC by a large margin. I don't think the "lost team" argument has much merit. Apple is still delivering very good performance improvements, and they have had substantial IPC improvements (just not on all possible workloads) since A14. I don't really see any momentum loss here.

What is this story about their "huge area cores" though? The A17 Pro P-cores are smaller than AMD's Zen4 and less than half the size of Intel's cores.

Competing is going to be next to impossible right now, because they'll be having a large stake in being ARM-based, while having no options but to increase area even further or raise freq.

I'm not quite sure what you are saying here. Apple's performance improvements have been solid and on time. And they seem to balance the IPC, performance, clock, and power consumption well. If you argue that they are in trouble unless they can improve the IPC, you would need to demonstrate that some other product can surpass their dominance in this segment. I am curious about the new Cortex cores, for example. If ARM managed to match the A17 pro performance on a similar power budget, that would lend your argument more credibility. Of course, we'd need to get some clear performance figures. ARM's slides gave me headache.

Just going back to Intel/AMD is going to be bad, and they can't improve their IPC at all, haven't for years. I'm going to laugh so hard if in a few years they're forced to buy QC, that would be the ultimate irony.

How would buying a CPU designer with a lower IPC help them increase their IPC?