[Bloomberg] Apple starting process to dump Intel in Macs

[Ajay]

Could be time for price negotiation with Intel. Or could be the real deal.

Great comment! Apple is known for pressuring suppliers; this could be a purposeful leak so that Apple can push for better terms. As an added bonus Apple can pit Intel 10nm against AMD “7” nm products for even more pressure. Sucks being Intel right now. The 10nm fiasco has really put them in a weaker position vis-a-vis margin protection.

 

[rainy]

Sucks being Intel right now. The 10nm fiasco has really put them in a weaker position vis-a-vis margin protection.

Poor guys!
They have just 63 percent margin, when AMD 35./s

 

[Ajay]

Poor guys!
They have just 63 percent margin, when AMD 35./s

Weaker going forward compared to the past. Intel is still a premium brand to consumers. OEM are probably questioning that to some extent, but with no end user ad campaign by AMD, brand loyalty will still be an Intel positive.

 

[Atari2600]

Don't forget that Apple also gets the benefit of no longer giving the PC market a leg up. Apple invented the MacBook Air, Intel repackages it as the Ultrabook and sells it to their competitors. Apple requested that Intel get the TDP and size down low enough to make the fanless MacBook possible, and Intel turns around and sells it to their competitors.

Wow, wait a sec.

So Intel does the development hard yards while Apple get exclusivity 'cos they asked for it?

Get real. If Apple didn't pay a hideous amount of R&D directly to Intel for it - then Intel would be daft to give them it exclusively.

 

[jpiniero]

Why would they not want design Semi-Custom GPU from AMD for their desktop computers?

Because they like doing it themselves.

 

[FIVR]

Looks like my idea of an AMD designed co-processor is in the works. Apple is abandoning Intel, not x86


https://twitter.com/BitsAndChipsEng/status/981135048429752322

 

[UsandThem]

Poor guys!
They have just 63 percent margin, when AMD 35./s

Which normally would be considered a great position to be in, except they had almost 88% of the CPU market as recently as Q4 2015.

https://www.pcworld.com/article/306...ng-itself-as-a-formidable-rival-to-intel.html

 

[jpiniero]

Looks like my idea of an AMD designed co-processor is in the works. Apple is abandoning Intel, not x86

https://twitter.com/BitsAndChipsEng/status/981135048429752322

B&C is pretty notorious AMD fanboys... they know nothing.

 

[scannall]

Looks like my idea of an AMD designed co-processor is in the works. Apple is abandoning Intel, not x86


https://twitter.com/BitsAndChipsEng/status/981135048429752322

Seems more like wild eyed speculation to me. Would it be cool? Sure. But that thread is just crazy talk, and not proof of anything.

 

[PeterScott]

Looks like my idea of an AMD designed co-processor is in the works. Apple is abandoning Intel, not x86


https://twitter.com/BitsAndChipsEng/status/981135048429752322

I am sorry, but just NO. They are misinterpreting (willfully?) what Bloomberg said.

Bloomberg did in fact say Apple is abandoning x86, even if they didn't say it one actual sentence in those exact words.

Here is one of many quotes where they mention going to ARM:

For Apple, the change would be a defining moment. Intel chips remain some of the only major processor components designed by others inside Apple’s product portfolio. Currently, all iPhones, iPads, Apple Watches, and Apple TVs use main processors designed by Apple and based on technology from Arm Holdings Plc. Moving to its own chips inside Macs would let Apple release new models on its own timelines, instead of relying on Intel’s processor roadmap.

It's pretty damn clear:

Intel x86 out, Apple ARM in.

They are talking about Abandoning Intel for their own in house ARM CPUs. Since Intel is Apples x86 source, and they are moving to Apple designed ARM, there really is no room for any other interpretation.

 

[deathBOB]

Looks like my idea of an AMD designed co-processor is in the works. Apple is abandoning Intel, not x86


https://twitter.com/BitsAndChipsEng/status/981135048429752322

Trade one unreliable supplier for weaker less reliable supplier? I don’t see it happening.

 

[Glo.]

Because they like doing it themselves.

Apple has capabilities to design iPad/iPhone Graphics.

They do NOT have capabilities to design desktop grade GPUs.

Vega/Titan Xp sized.

 

[scannall]

Apple has capabilities to design iPad/iPhone Graphics.

They do NOT have capabilities to design desktop grade GPUs.

Vega/Titan Xp sized.

This makes no sense at all.

 

[Glo.]

Trade one unreliable supplier for weaker less reliable supplier? I don’t see it happening.

Less reliable? Semi-Custom business is different than getting "normal" SKUs that are widely available.

Semi-Custom business means that Apple comes to AMD with specific needs, and they work not with GloFo, but with ANY fabs that are willing to cooperate with Apple: Samsung, TSMC, Intel, GloFo, whatever.

It has nothing to do with reliability, but everything to do with what Apple needs(Lets say something custom based on Zen architecure, and next gen. GPU arch, for graphics). For AMD its bad business, apart from PR. For Apple its best business because it helps them maximize profits, and not rely on Intel's roadmaps, and get what is best for their needs.

 

[Glo.]

This makes no sense at all.

Really? Scale a GPU from small SoC, that has less than 100 mm2 die, and includes EVERYTHING computer has: CPU, GPU, RAM, other coprocessors, etc, to single discreet GPU, that would have over 450 mm2 die size, for example.

 

[coercitiv]

Apple has already stated that they are immediately re-designing their chips to be immune to these attacks.

So did Intel.

When is intel going to have chips that are also immune? Apple is slated to release a fixed A12 in November 2018.

Q2 2018, same as Apple

Intel isn't going to fix Spectre or Meltdown in hardware anytime soon.

This is pure FUD.

 

[scannall]

Really? Scale a GPU from small SoC, that has less than 100 mm2 die, and includes EVERYTHING computer has: CPU, GPU, RAM, other coprocessors, etc, to single discreet GPU, that would have over 450 mm2 die size, for example.

GPU's scale up very well. So I'm not seeing a problem here.

 

[moinmoin]

AMD's semi custom business would be perfect for Apple. Only problem with that rumor (that started in 2015) is that AMD recently restructured its business with the semi custom unit being moved to RTG. They would not do that if they had a major semi custom project going on with Apple containing x86 IP.

 

[bsp2020]

My guess is that, if AMD is involved, Apple would license AMD64 & Zen micro-architecture and implement it themselves. AMD has licensed AMD64/Zen before (THATIC). So it's not a new idea for AMD. Also, for Apple, it won't make any difference whether they use ARM ISA or AMD64 ISA as long as they design and manufacture their own chips. I think software compatibility is much more important to them then ISA religion. It may even make sense to have both ARM and AMD64 cores in a single chip and run both iOS and macOS using their native ISA.

When they transition from PowerPC to Intel, Intel had performance edge that smoothed out the transition period. I don't see Apple chip providing such performance edge over Intel in the near future. Even the best ARM server chips are not as fast as Intel core per core and single core performance still matters for desktop use. If Apple is ambitious, maybe they will finish what AMD started (one micro-architecture that can handle both ARM ISA and AMD64 ISA).

 

[FIVR]

So did Intel.


Q2 2018, same as Apple


This is pure FUD.

These are facts:

Intel has provided no hardware that is immune to spectre and meltdown

Intel has not claimed it will provide any hardware that is immune to all variants of spectre.


All intel has claimed, is that it will fix meltdown and the first variant of spectre. They refuse to address the last variant because the performance improvements from allowing speculative executions is too high (or they are simply lazy/pick one).

See here in this Intel-sponsored article from Tech crunch. Straight from the horses mouth:

There are actually three semi-related bugs here: Spectre is variants 1 and 2; then there’s variant 3, which researchers dubbed Meltdown. Variant 1 is arguably the most difficult of them all to fix, and as such Intel doesn’t have a hardware solution for it yet — but variants 2 and 3 it has in the bag.

https://techcrunch.com/2018/03/15/i...s-for-spectre-and-meltdown-on-upcoming-chips/

 

[PeterScott]

See here in this Intel-sponsored article from Tech crunch. Straight from the horses mouth:

https://techcrunch.com/2018/03/15/i...s-for-spectre-and-meltdown-on-upcoming-chips/

You are really dragging this dead horse into every story. This has nothing to do with Apple Switching Mac to ARM.

Where exactly did Apple say there chips would be Spectre Immune? I would like to see the actual wording, not your paraphrasing.

 

[maddie]

I thought overall desktop & laptop sales where in decline or a holding pattern? Seems kinda odd for Apple to be putting resources into moving away from Intel and into their own processors.

It's exactly when a market begins to stagnate that vertical integration accelerates. Got to keep those profits and margins growing. Apple reaching saturation with the present product lines almost forces this move, courtesy of the stock traders.

Bit of risk however.

 

[wahdangun]

GPU's scale up very well. So I'm not seeing a problem here.

No, if that was easy then why intel struggling to launch it's own gpu design? They even scrapped their larabe platform even when they waste RnD cost, and that's why there's only 2 major player in gpu, the barrier is hard and expensive.

And in the cpu world IPC is not everything, it's combination of IPC, clock, and price. And you can't get all at once.

 

[itsmydamnation]

GPU's scale up very well. So I'm not seeing a problem here.

GPU do have problems scaling, they have the same kinds of problems as scaling CPU count does, you just dont understand how GPU's work. If you wanted you could relate a GPU to Ryzen, a Shader engine (a CCX) has X number compute units ( cores), each compute unit has SIMD units within(just like CPU's). increasing any one of these has a scaling effect all the way up the stack. As it stands now AMD are "stuck" at 4 Shader engines and NV's equivalent is 6. This will be because this is the point they join the shared L2 along with all the other stuff that has to join the L2 like ROP's, MTU's this interconnect point is hard and very expensive to scale, as far as im aware they are full or partial meshs, not very nice for scale-abilty. Increasing unit counts within a SE or SIMD within a CU will just lower performance per unit unless you also scale your read/write bandwidth and outstanding request queues all the way up to main memory.

As a result unit count increases has slowed dramatically and per unit performance has been the target for increased performance. if your rolling with a DSBR this is even harder then a IMR (like AMD/NV) because you need a lot more storage space on your chip with tighter integration between separate GPU functional units .

There is no doubt Apple has the budget to do it ( but i doubt they have the revenue in the target market to make it worth while) but it will be a 3-4 generation 6-8 year project to get on par with AMD/NV, just like it was on the CPU side.

 

[beginner99]

2020 is enough time for Apple to design their own. Intel and AMD are pretty much standing still compared to Apple. Since 2011, Intel went from Sandy Bridge to Coffee Lake (~2x), while Apple went from A5 to A11 (~10x)

Yeah if you start with 1 horse is easy to get to 10 hp. however if you start with 300hp it's a lot harder or shall I say cost prohibitive to hit 3000hp. Relative comparisons sometimes really are meaningless. If your 10% gain means a higher absolute performance gain than the competitor doubling in performance, who actually is the winner? Hint: the gap increased.

Note: I'm not defending Intel just saying it's always easier to have huge relative gains if you are starting from a very low baseline.

It's not really surprising. Why keep paying Intel margins for stagnant architecture and stagnant process?

Why invest into chip design for a marginal market? Mac market is tiny compared to iPhone and iPad meaning you now have to divide the chip design effort over a rather small amount of devices. Not worth it. Probably cheaper to put pressure on Intel and I'm sure Apple does so you know, by spreading rumors like this.