All iPads Chipset "innovation"

[Italianblend]

So, tech companies like apple improve their products every year. Apple consistently and without fail put out a better chip to power it's mobile devices every year. Usually twice a year like A8 and A8x. Next year will A9 and so on...

For example, in the keynote, the apple guy said last year's chip had a billion transistors and this year's model has 3 billion. This seems like a big step. Did the technology exist last year to make 3 billion? Or did we discover some innovative way to add transistors to a chip sometime in the past year?

I know nothing about how chips work or are made. Here are a few essay questions for people who do.

1) Does the technology exist today (October of 2014) to produce the chip that apple will release in the iPad model of 2017?

1y) if you answered yes, what is your reasoning behind the fact that the technology is being "held back," meaning the latest tech isn't in the hands of the public? Is it simply too expensive to create? Or, are tech companies ensuring future customers by ensuring that next year's technology is a sure thing and gives people reason to upgrade?

1n) if you answered no, then how do you account for the fact that apple is without fail producing a better chip each year?

 

[cycledance]

So, tech companies like apple improve their products every year. Apple consistently and without fail put out a better chip to power it's mobile devices every year. Usually twice a year like A8 and A8x. Next year will A9 and so on...

For example, in the keynote, the apple guy said last year's chip had a billion transistors and this year's model has 3 billion. This seems like a big step. Did the technology exist last year to make 3 billion? Or did we discover some innovative way to add transistors to a chip sometime in the past year?

I know nothing about how chips work or are made. Here are a few essay questions for people who do.

1) Does the technology exist today (October of 2014) to produce the chip that apple will release in the iPad model of 2017?

1y) if you answered yes, what is your reasoning behind the fact that the technology is being "held back," meaning the latest tech isn't in the hands of the public? Is it simply too expensive to create? Or, are tech companies ensuring future customers by ensuring that next year's technology is a sure thing and gives people reason to upgrade?

1n) if you answered no, then how do you account for the fact that apple is without fail producing a better chip each year?

1) possibly very early prototypes. concepts for sure.
1y) needs more testing. needs new mass production facilities.

 

[Italianblend]

I'm not buying that. I think technology is slowed to maximize profits.

 

[Chupa Chupa]

I'm not buying that. I think technology is slowed to maximize profits.

Sure, that's why Intel pushed back availability of the new Broadwell chips... to maximize profits by not selling as many chips as possible and pissing off its major customers.

Chip engineering isn't as easy as you suggest, but Moore's Law (or the rough concept) goes on. It's all very incremental and research and testing of one generation helps in engineering the next. It's not that they invent the Holy Grail of chips then dumb it down adding back a little power each year. If you look as the generational changes they are each distinct.

 

[Cyrax]

A7 was 28nm, A8 is 20nm. Die shrink lets you cram more transistors onto the CPU while not increasing the size.

The 2017 chip is on the roadmap, but no, the technology does not exist today.

 

[ZBoater]

Ever heard of Moore's Law?

http://en.m.wikipedia.org/wiki/Moore's_law

It's like Murphy's Law, but better.

 

[rrm74001]

I know nothing about how chips work or are made.

Well that sums up this thread pretty well.

 

[TechGod]

I'm not buying that. I think technology is slowed to maximize profits.

You just admitted that you don't know how chip manufacturing works then you turn around and tell the guy that told you about chip manufacturing, that he is wrong

No tech isn't held back for profits. You really don't understand this so don't make assumptions.

Chip production is no where near as easy as you think it is.

 

[pmau]

... and most of Chip design is ruled by physics
You cannot simply draw a new chip on paper and say "Aye, let's double the amount of registers, internal cache and add more memory address lines".

The new chips are possible because the structures get smaller.

The problem that a signal cannot reach the other side of a chip fast enough has brought us multiple cores. You just create two signals that travel half the surface to get a result.

Of course you can design a chip that is faster, but you have to improve the process of building it without side effects and very small margin of error.

This is why modern chips are completely designed by programs that know the physical limitations of signal decay and timing based on distance / latency.

 

[KeithJenner]

I'm not buying that. I think technology is slowed to maximize profits.

Impressive, you got all the way to post 3 in the thread before abandoning the whole basis of your opening post.

If you really wanted to just state your opinion then it would have been easier to have got it out of the way In post 1.

 

[Solver]

...technology is being "held back"?

Yes, Apple is holding back from the public in house versions of iOS 8.2 and may even be holding back an early alpha version of iOS 9. Time for the people to revolt!
(sarcasm)

 

[justin216]

They almost certainly have rough plans for the next 5-10 years, but most of it isn't possible yet. The reason we can't jump from 20nm CPUs we have now in the A8 to, say, 5nm is because for every reduction in size, the laws of physics start compounding. Heat, ability for bits to be read/written, keeping electrical signals separate, etc. It all adds up.

So, you do it a step at a time. You go from 28nm to 20nm as soon as the aforementioned hurdles are overcame. You then refine the process to increase yield and lower costs. Apple tends to stay on the same rough size for two years, although mid-late second year they do tend to die shrink (i.e., A5 went from 45nm to 32nm over time).

You also have to factor in that there aren't many chip fabricators operating at these small sizes -- literally a handful. They need good yields from each wafer to make the millions of devices required, and it takes time to get the chip designs right.

You can bet Apple is already far into development on the A9 (likely late stage prototyping it), and have early test designs for the A10.

While we see these new things annually, they're realistically in development for several years. Even in years when they can't die shrink, they beef up the GPU, or improve the logic, or add more cache, etc. A CPU platform is ever-evolving.

 

[Italianblend]

What was the big discovery last year that allowed apple to go from 1 billion to 3 billion transistors?

 

[Donka]

What was the big discovery last year that allowed apple to go from 1 billion to 3 billion transistors?

No discovery, it is purely a manufacturing process. The companies manufacturing the Ax processors for Apple have moved to a smaller manufacturing process (die size not assembly line!). This lets them fit more transistors into the same amount of physical space or have the same number of transistors but use up less physical space and less power.
The A8x achieves it's transistor count increase by switching to a smaller die and possibly even increasing the size of the chip slightly if required.

Die shrinks are a gradual process than needs to be perfected as the lithographic process used is improved which is why we see gradual improvements in this regard on a yearly basis.

 

[Italianblend]

Okay, then let me rephrase my question.

Does Apple have the technology (or manufacturing process) today to produce the chip that will be put into next year's iphone and/or ipad?

 

[Chupa Chupa]

Okay, then let me rephrase my question.

Does Apple have the technology (or manufacturing process) today to produce the chip that will be put into next year's iphone and/or ipad?

Of course Apple has already designed the A9. It was already reported this summer. Samsung will be ready to ramp up production next year. TSMC, which manufactured the bulk of the A8 chips does not have the technology to make the A9 which is based on a 14nm design.

It's highly likely the A10 and A11 design process and testing is ongoing right now. Conception to production takes well more than a year. Companies have many years of future potential components in the R&D pipeline. They don't just appear by spreading magic dust. A lot of testing happens before it's ready for market.

Chip manufacturers (Apple doesn't manufacture chips, it only designs them based on what current technology allows) routinely announce their roadmap for the next year or two. But just because a chip is announced for production sometime in the future doesn't mean it's ready for production today, or that the production goals will be met. Again, look at Intel's Broadwell chip and it's delay.

 

[Piggie]

I am only guessing, but isn't the BIG number jump simply just going to br mainly a doubling of memory?

Nothing exciting, just a big fat chunk of memory transistors, rather than a big fat junk of clever new processing areas?

I'm just stabbing in the dark here, but could it mostly be that?
Just basically a extra large chunk of empty space for memory had been added?

----------

(Apple doesn't manufacture chips, it only designs them based on what current technology allows)

Isn't that slightly stretching the truth?

Is that like me saying I make houses, when what I'm doing it re positioning pre made lego bricks constructed and made by others, and I'm just really changing the layout of their bricks, which I bought the rights to use, and perhaps modifying a few of the bricks into some slightly better shapes for me?

 

[Chupa Chupa]

Isn't that slightly stretching the truth?

Is that like me saying I make houses, when what I'm doing it re positioning pre made lego bricks constructed and made by others, and I'm just really changing the layout of their bricks, which I bought the rights to use, and perhaps modifying a few of the bricks into some slightly better shapes for me?

How is that stretching the truth? Please explain intellectually without a convoluted metaphor.

Samsung and TSMC manufacture the A series chips for Apple based on those respective companies manufacturing technologies. That is the un-stretched fact. Apple is not a chipmaker, it's a CE company. As such it has the option of buying chips off the shelf, or designing a custom one. Either way it has to go to a chipmaker like Samsung and TSMC.

 

[Piggie]

How is that stretching the truth? Please explain intellectually without a convoluted metaphor.

Samsung and TSMC manufacture the A series chips for Apple based on those respective companies manufacturing technologies. That is the un-stretched fact. Apple is not a chipmaker, it's a CE company. As such it has the option of buying chips off the shelf, or designing a custom one. Either way it has to go to a chipmaker like Samsung and TSMC.

Ok, well, if you said to me you made a chip, then I would think by that you mean that you started with a totally blank sheet of paper, and using some, I presume software for doing this, designed from scratch, from nothing at all. No starting point your chip.

That's what it is, in the same way if you told me you painted a picture, that means you took a blank white canvas, some brushes, some paints, and painted it.

Please correct me here, but Apple and others use ARM from the UK, which started out back in the 1980's designing the 1st ARM RISC Chips for the Acorn Archimedes line of home computers.

ARM offers two types of licenses to its clients: a processor license and an architecture license.

I have no doubt that Apple Tweak, Adjust, Improve, change, modify etc etc the Arm chip/design into what they want, and make it better to suit Apple's devices.

But, if that is the case, to me that's not "Making the chip"
Licencing someone else's invention/product is not the same as blank sheet of paper making anything.

If this is wrong factually then please correct me.

At the moment I see Apple's chips as modified ARM chips, I don't see them as 100% from the ground up Apple design, as if they were then Apple would not need a licence from Arm or anyone as it would be all their work.

Perhaps you can cleanly clarify this point for me?

If I licence bits from you that YOU invented over decades from scratch, and I modify your bits, making them better and more suitable for my needs, then I can't in all honesty say it's MINE I made it.

 

[Chupa Chupa]

Ok, well, if you said to me you made a chip, then I would think by that you mean that you started with a totally blank sheet of paper, and using some, I presume software for doing this, designed from scratch, from nothing at all. No starting point your chip.

That's what it is, in the same way if you told me you painted a picture, that means you took a blank white canvas, some brushes, some paints, and painted it.

Please correct me here, but Apple and others use ARM from the UK, which started out back in the 1980's designing the 1st ARM RISC Chips for the Acorn Archimedes line of home computers.

ARM offers two types of licenses to its clients: a processor license and an architecture license.

I have no doubt that Apple Tweak, Adjust, Improve, change, modify etc etc the Arm chip/design into what they want, and make it better to suit Apple's devices.

But, if that is the case, to me that's not "Making the chip"
Licencing someone else's invention/product is not the same as blank sheet of paper making anything.

If this is wrong factually then please correct me.

At the moment I see Apple's chips as modified ARM chips, I don't see them as 100% from the ground up Apple design, as if they were then Apple would not need a licence from Arm or anyone as it would be all their work.

Perhaps you can cleanly clarify this point for me?

If I licence bits from you that YOU invented over decades from scratch, and I modify your bits, making them better and more suitable for my needs, then I can't in all honesty say it's MINE I made it.

OK, I see the disconnect here. English must not be your native language so when you original quoted my post which said "Apple doesn't manufacture chips," you must have understood "manufacture" to mean create from ground up, rather than the physical production of a good.

I though I was fairly clear when I said Samsung and TMSC make the chips for Apple, but language barriers and all can trip up even the best and brightest.

 

[Piggie]

OK, I see the disconnect here. English must not be your native language so when you original quoted my post which said "Apple doesn't manufacture chips," you must have understood "manufacture" to mean create from ground up, rather than the physical production of a good.

I though I was fairly clear when I said Samsung and TMSC make the chips for Apple, but language barriers and all can trip up even the best and brightest.

Indeed, a misunderstanding.

Of course Apple does not Manufacture the chips as it's not a manufacturing company. I took the meaning wrong, when someone said make the chips
No probs

 

[Gincoma]

Its mind boggling how these chips are made, how can you fit 3 billion of anything on something so small is it done 1 resistor at a time in lighting speed, its amazing how technology evolves.

 

[leonk1]

Silicon is rapidly approaching a dead end named physics. We are almost to the point where cramming more transistors onto a die simply will not work.

Personally I thought we would already have new processor materials by now, but it appears that Intel and others are sticking with silicon for the next several years. Processor improvements are going to slow to a crawl.

 

[Erasmus]

Okay, then let me rephrase my question.

Does Apple have the technology (or manufacturing process) today to produce the chip that will be put into next year's iphone and/or ipad?

Maybe.

If it does, it would likely cost them thousands of dollars each to make. Maybe less if they print it at the present 20nm process, but that would seriously impact energy use.

2017? No. I would be flabbergasted if there exists any facilities anywhere in the world capable of large-scale <10nm lithography.