iPad's Apple A4 CPU is feature-stripped ARM Cortex-A8 - report - Page 2

Disappointing...Its still faster than the iPhone processor due to clockspeed increases, but an The A9's provide an enormous benefit, they are out-of-order processors for one, which is a MASSIVE benefit in itself, and for two they are either dual or quad cores. I cant remember who's specific test it was, but the A9 kept toe-to-toe with Intel's Atom (pinetrail) while consuming so much less power. The A9 would have been a screamer, the A8 is meh.



Apple will still market it as the god of all mobile devices though, which has me worried. I've already heard people say they want to replace their laptops with one, thinking it has all the same functionality. Those people are in for a big let-down.

It may be so or --- absolutely evenly --- it may not, too.
The fact "tech" blogger has managed to figure out block diagrams and I/O nomenclature of modern SoC (Freescale i.MX51) doesn't necessarily mean anything to A4 processor.

Unfortunately, many articles are repeating this incorrect information. The original article which is being quoted, said that it could cost $1 billion for a company to create a new processor from SCRATCH. That is, from no current design. This is not what Apple is doing. They are doing what everyone else is doing in the mobile space that uses ARM's. They license a processor chip, and modify it. The cost is well below $1 billion. It may cost as little as a few tens of millions.

Entirely agreed.
Pundits fail to understand the basic theories of corporate strategy that are clearly observable in Apple's businesses.
Innovative products like this are never intended to be a replacement of any preexisting solution.
The perspective is flawed.

Re #1:
I am not convinced. It does not make sense to me that Apple would go to all the effort of producing its own chip for so little gain. In the end, these are still only rumors.

Additionally, we have the few limited reports that system is really fast and responsive. Mossberg, for example, said "wicked fast." This indicates that there is a lot more going on than what the Ars Tech. article proclaims.

Personally, I see no reason to doubt the quote in #2.

If this article is correct, then it's actually a very good sign. We sometimes tend to be too fascinated with the latest hardware, and forget about the software running on the things.

Jon Stokes thought the iPad must be running the latest Cortex A9 dual core at 1GHz to get the speed and smoothness it had. If it uses the same chip as the 3GS, though faster, it means that Apple has gotten the OS to run VERY efficiently on this chip.

This is GREAT news! Much more important that NEEDING a much more powerful chip to do so. It means that when Apple has a much more powerful chip, this thing, and more importantly, its applications, will fly!

Why does the iPad need multiple cores? Why does it need an out-of-order processor? Those technologies might be "cool" but are they the right tradeoffs for the iPad? Honestly, I don't know, but I can imagine that they might not be.

The lack of multitasking for 3rd party apps might not be a temporary restriction -- it might be a fundamental design decision on Apple's part that will persist for quite some time. If so, then the need for multiple cores is greatly reduced.

What should be liked about AI's article is Sam has finally started to name sources. This is good thing; it happened to be worse before.
But in this case the source is just shameless and irresponsible speculation.

As is well known, the iPhone is multitasking all of the time.

What you are talking about is not multitasking, but the running of two or more apps at the same time, mostly as background processes. Apple does do that now, it just doesn't yet allow apps that aren't in the basic set to do so.

The cpu doesn't run the display, as is also well known. An Imagination chip does that, just as GPU's do in our other machines, so and other concurrently running apps will have no effect upon that either. And, as I've already pointed out, the device does do that already, so obviously, it works well.

Lack of vision my friend. You're putting this in a box before you even know much about it.

Sigh! It's going to become boring to continue to have to always point this out, but Apple didn't pay more than a few percent of a Billion dollars to "develop" this chip.

The original article which is continually being quoted out of context here, said that it could cost $1 billion to develop a NEW cpu chip from SCRATCH. Apple certainly isn't doing that here. They're are licensing an ARM chip that is closest to what they want, and making some fairly minor modifications to it.

I wish we can lay that $1 billion number to rest.

So one year old processors can't do multitasking?

Everyone needs to understand that the lack of multitasking is NOT because these iPhones/iPads don't have enough CPU. It's for other usability reasons, like:

-Power management (keeping background apps from using all your battery)
-Network management (what if background apps ping servers all day long? This is also a power management issue)
-App management. Apple "famously" mocked Windows Mobile task manager. The moment you have multitasking you add the complexity of having to have the user manage know about it, and manage it. How do background apps notify the user that *something* has happened? How do users kill unresponsive background apps?

My first PC was a 486 running at 66Mhz. It did multitasking just fine...

Apple probably will introduce multitasking at some point on these devices, but it likely won't be multitasking as you see in general purpose OSs. I suspect it will be a limited form and applications will have to be coded to deal with it, rather than just assume that they're running all the time. They'll be limited in CPU, network, and UI resources.

I would think it's unlikely they stripped that out. Why bother? It wouldn't save much power. And they may want it for a later model. It might result in a slightly higher yield, and lower cost. But, there would need to be a significant percentage of the chips area involved for that to be helpful.

Stripping out anything else that isn't needed would result in a smaller and cheaper chip, assuming that the chips's area was also reduced, rather than just leaving blank areas on the chips surface. But even so, less circuitry would result in less heat, less power, and better yield. Not a bad tart.

Can we even trust Ars?
Apple didn't disclose anything.

You're missing something here.

Foryears, the Palmphones used weaker processors than didWindows phones, but performed much better. Yes, its true they didn't have true multasking, which the Win Mobile phones had to some degree. But the difference in performance was much wider than that one thing alone could account for. I still have my old Teo 700p. It's blazingly fast at whatever it does. No delays whatsoever. Yet, the 700w model, with a faster processor was far slower at everything.

What we have here, if this aticle is correct, is an OS and software that can operate amazingly well on this hardware, and that says something important.

One this it says is that Apple's battery life claims, which are much better than even small Windows tablets gives it a significant avantahe. We can also look at the JooJoo. It have less than half the battery life using Android, though it does have a 12" screen. The Acchos 9" with Windows 7Starter has even less battery life than that. And it uses a weak tom chip.

This looks very good for the future direction of the line.

My 3G is struggling to run its software at times. But it manages pretty well almost all the time. I had read that Apple used every processor cycle in these earlier devices. That's why the 3Gs screams, even surpassing the Nexus One in many tasks, even though that uses a newer chip.

I think Apple is going the right way here.

Gates once said that his job was to add all the features they could think of, and let the computer manufacturers figure out how to make it run fast enough. Apple seems to be going the other way. Make the OS and software work efficiently, so that the hardware designers have an easier job of it.

I like that idea.

All good points. I would just add that human "multi-tasking" isn't really multi-tasking, it's task switching. I cannot write an e-mail and read a blog at the same time. I physically cannot do it. A multi-core computer, on the other hand, can literally do two things at once, for example encode video while running a simulation.

For the iPad/iPhone, the key is to enable human-style multitasking, which is to say rapid task switching. I need to be able to switch between my e-mail program and the web browser quickly and without losing the state of the respective apps. The iPhone isn't perfect by this measure, but improving this does not require actual multitasking of apps. It just requires that the state of apps be saved on exit and then perfectly restored when relaunched, and that these things happen very quickly. I bet we see this in iOS 4.

As for actual computer multi-tasking, the need for it on the iPhone/iPad is very limited, and already exists for the most part. The only thing that I cannot do on my iPhone that I would like to do, and which does require multi-tasking, is to run Pandora in the background (I can only listen to Apple music in the background). So if Apple were to allow a rare exception to the "no 3rd party background tasks" rule, and allow them when the developer shows that it really is necessary and that it really does work ok, then that problem would be solved, too.

But I will never be compressing video or running simulations on my iPhone, so the need for unfettered 3rd party app multitasking just isn't there. And this means that there is very limited need for a multi-core CPU, too.

We know it cost more than a few percent since buying PA Semi cost 28% just to play the game. I have no ida exactly how much it cost and clearly referred to the $1B a a rumour, but I also don't think it's impossible to have cost that much.



I think so. The article seems well balanced and logical. Of course, none of this is set in stone until we get to see the breakdown. it'll be interesting to see what was added and removed.

I suppose it's interesting to speculate about but odds are no-one outside Apple (& Samsung?) know exactly what the A4 is and what it's derived from. All this speculation without direct knowledge or facts seems pretty meaningless.

Feedback from MacWorld from people who actually used an iPad said it appeared very very responsive. That's all that matters. Apple's approach is to let go of focusing on the hardware on this device. This is all about software and usability. The iPad is directed at consumers and for business applications where people just want a tool to get the job done without the support headaches of a PC/notebook - and could care less what generation processor / graphics chip / OS / etc. is inside...

Yeah. And if it appears to be Cortex-A9, I personally bother to make laughing bags out of trolls. And I can.

If Apple gets rid of MacBook to replace it with iPad i think it will lose quite a few customers. Of course they can just bring down the pro line a little bit (it's already quite inexpensive with the 13 incher) but I see a lot of MacBooks at school and only a few Macbook pros. I doubt iPad can replace MacBook, especially not with its current iteration. I'm not saying that it never will, but Apple needs to take a breather and think twice before getting rid of the MacBook.

As far as the Pad running an older processor - kinda makes sense, since apple would have started the project a while ago. This does not make this an A8 processor though, since its a fork and could have been greatly improved by apple's engineers. Since iPad won't multitask, it does not need a multicore processor, and the OS can be simpler and smaller without the need to support multicore processors.

Jon Stokes is a well respected technology writer. He's also a system design expert.

People have been getting too excited about PA Risc. Some of us have been saying, from the beginning, that it will take some time before they will be able to come up with anything significantly different.

Apple has only owned them for about 20 months. It takes a good six months for a company to change direction they way they did. The few people there wih ARM experience hadb;r worked on ARM;s for years. It would take them a while to get back up to speed. Then a group of top engineers left the compant to form their own because they eren't happy about the amount of stock Apple offered the, That had to hurt as well. They were busy hiring ne peopl duringthat time, and still are. Then they would have needed to decide what they wanted to do first, and decided on which ARM chips they had to license to to that. Then they would have to study those chip designs along with some of ARM Holding's engineers. Then they could sit down and start to make block changes in the designs. Only after all that was done, could they start the actual design process. After several months of that, they needed tapeouts, to make sure it would work, and to debug the results. Usually there are three tapeouts, rarely two, if everything is just dandy. Then they need to have a test run for samples to make sure that masks work properly. Then after that is finished, they need to make a longer test run to check initial viability of the wafer projections.

After all of that is successful, they have to make some pre-production runs to get the process running properly, so they could get a decent percentage out of the chips, known as yield.

It's not likely that with the time Apple has had, that they could have done very much with a chip. don't forget that these chips would have needed to be in production for at least a couple of months already.

Jon Stokes is well respected. What he says makes sense. Sometimes the articles here are a bit overblown, but the technical articles in Ars are also well respected. He's also saying that it looks as though it's true, but he's not saying it's 100% yet. I would tend to go with what he says unless shown differently.

The cost to buy PA Risc should not be added into that cost. It's the cost to develop the chip itself that's being talked about. he original article was talking about any chip manufacturer. Even those with facilities. Please don't get into a semantics game.

There is no way it could have cost more than a fraction of that. Apple isn't developing a chip. They're modifying a well designed part. If Jon is right, the most they seem to have done with this first part was to cut out pieces, rather than to do much in the addition area.

We'll just have to see what was done.

Right now, considering the thrust of the article, the trolls are the ones insisting that it's a Cortex A9, not the other way around.

No, no. Trolls are trying to mess up as much nonsense as they can to divert the attention from few reasonable comments in the topic.
As for what A4 processor really is, it's reasonable to wait for some reliable evidence and trustful explanations from processor's maker.

If Apple didn't intend to replace the MacBook line with iPads, they wouldn't have phased out all the successfully selling MacBooks over the last year and just leave one.

The iPad has a lower barrier to entry than the $999 MacBook, also the accessories are optional.

Because the price is lower, it's going to sell more and thus should gain rapid adoption. But it won't beat a netbook, which is what Apple should have produced.

I think I'm going to lean in the A9 direction if Steve Jobs said this chip "screams". Because apparently everyone is saying the A8 is "meh".

PA Risc?

PA Semi.

PA RISC is HP's RISC processor family.

Secondly, you don't invest $1 Billion in R&D on legacy ARM CPUs.

Unless you work for Apple you have no solid evidence that Apple intends to replace anything. In fact the iPad is a "companion device" meaning that it requires a host computer to sync to. You cannot buy an iPad and run in sans computer running OS X or Windows. So based on the information we have today the iPad is not going to suffice as a wholly separate product on its own island.

You keep infusing every iPad article with your FUD which is so poorly supported with actual fact or logical presumptio it's getting really old. You may be right, but right now the data out here doesn't support your hypothesis.

I disagree on all parts.

The iPad is not designed to replace a notebook running a desktop OS.
They streamlined the MacBook lineup, they aren't phasing it out.
They clearly priced the top of the line iPad lower than the MacBook.
With the iPad starting at half the price of a MacBook they'd have to get the double the net profit of a MacBook to make profit.
The iPad is designed and advertised as an accessory to a PC, not a replacement for it.
Netbooks are their own worst enemy. They try to be multi-function pocket knife and while they might do well in a pinch or be fine for kids they are can't do any one thing very well. We'll likely see Chrome OS or Android OS with a new UI be common on future netbooks because of this.

Apple aren't talking about the inner workings of the A4 and they're not going to. Remember they're a platform company: they don't want you thinking about how the hidden inner workings of the iPad does or does not stack up to the competition, all they want you to think about is how great a time you're having using it! Will it feel fast enough? Does the battery last for as long as you need before recharging? Is the application experience a good one and does it meet your needs? Are you enjoying yourself? If the answer turns out to be yes to all those questions, then you're not going to care what's inside the iPad, and Apple will have done their job well.

That is disappointing if true. There is something fun about bragging rights, and I would really like to see their custom silicone kick butt over everything else, but it really doesn't matter - if the iPad does what it's supposed to do and does it well, who cares what CPU/GPU it has? I would say it's not as big of a deal what kind of processors special purpose devices have compared to general computing devices.

That's the thing, it doesn't mean it won't kick butt. They are using a custom chip with a custom OS designed for each other, which means the product could very well be faster than anything else on the market. The resulting product is the only thing that matters.

Remember that Apple hired two specialists, Mark Papermaster from IBM and Bob Drebin from AMD (I think). And that already back in 9/2008, Wei-Han Lien (Apple senior manager who came from PA Semi) had posted on LinkedIn that he was working on a new ARM chip for the next generation of iPhone (which could easily be also for this iPad).

I think it was Cook who also specifically pointed out last week that the key to better processing and better battery life was removing things that weren't needed in a customized SoC. So whether Apple started with an A8 or A9 might not be that important for this first iPad.

But also note that the iPad has room for a much larger battery.

You have to look at prices. Apple didn't phase out the MacBook line; it converted them into 13" MacBook Pros with better specs for the same prices.


An Apple netbook would not have been priced at $499. But regardless, we'll see if you're really smarter than Steve Jobs and his A-team of engineers and marketers.

He-he. A8/A9 dilemma has sure got nothing to do with the quality of Apple's product. Apple only states so far one thing: their processor is outstanding, when it comes to energy consumption. And we have no reasons to doubt what Apple says. We've equally got no right to judge whether they managed to reach that efficiency elegant.

But it's not the first time Ars is spreading suspicious speculations instead of exact information. It's Ars' credibility, which now looks questionable.

This is incorrect. Despite the fact that the OS running on the iPad and the OS running on the iPhone are both called iPhone OS, the version running on the iPad is a (likely non-strict) superset of the version running on the iPhone -- i.e., it includes the iPhone functionality as well as additional functionality not available on the iPhone.

The fact that the iPad runs 3.2 is IMO due to an anticipated release in regard to iOS 4.0. I expect apple to run the exact same os on both the iPad and iPhone, but with different functionalities enabled for the iPad once this version will come out this summer, and hopefully announced later this month.

I don't think they will end up with to iOS lines.

It's odd that they call them both iPhone OS because they are significantly different. Not just the UI but core technologies that make it very unique. That is why iPad apps can't run on the iPhone/Touch and why you can't use iPhone OS 3.2 SDK to make iPhone or Touch apps. They really are quit different in spite of the nomenclature.