Apple's iPad - The AnandTech Review
by Anand Lal Shimpi, Brian Klug & Vivek Gowri on April 7, 2010 9:39 PM EST- Posted in
- Smartphones
- Mac
- Apple
- iPad
- Mobile
The Silicon - Meet the A4
Before the first iPad was ever unboxed, photos had already leaked of its dissection. Admittedly, I was relieved as it meant I didn’t have to risk destroying mine to figure out how to get inside the chassis. Why was getting inside Apple’s new tablet such a big deal? Because it’s using Apple’s brand new SoC.
It’s called the A4 because it is the 4th generation SoC designed by Apple. That would make the silicon in the iPhone 3GS the A3, the A2 would be what’s in the iPhone and perhaps A1 was an older iPod chip? Either way, the A4 was the first we heard of Apple’s naming system.
For months leading up to the iPad announcement everyone assumed that Apple’s acquisition of PA Semi meant that the company was going to ditch Samsung and come forth with its own killer SoC. Would it be an out-of-order Cortex A9 design? Or would it be truly custom, similar to what Marvell or Qualcomm have done with their Armada and Snapdragon lines?
As with everything involving the iPad, rumors of the device’s silicon ran rampant. But the idea of Apple doing their own microprocessor architecture for the iPad was disappointing to me. To understand why, we have to look at Apple COO, Tim Cook’s recent statements on why Apple has been successful:
“We are the most focused company that I know of or have read of or have any knowledge of. We say no to good ideas every day. We say no to great ideas in order to keep the amount of things we focus on very small in number so that we can put enormous energy behind the ones we do choose.”
Tim was talking about products there, but I believe his statement is applicable to the A4 as well. Apple is not a microprocessor company, nor does Apple want to toss its hat in with the likes of Intel, NVIDIA, Qualcomm and TI as an SoC maker. History has shown us that the only way to be a successful microprocessor company is to be able to subsidize the high cost of designing a powerful architecture over an extremely large install base. That’s why x86 survived, and it’s why the ARM business model works.
Designing high performance SoCs just for use in the iPad and iPhone just doesn’t make sense. In the short term, perhaps, but in the long run it would mean that Apple would have to grow the microprocessor side of its business considerably. That means tons of engineers, more resources that aren’t product focused, and honestly re-inventing the wheel a lot.
The fact that the A4 appears to be little more than a 45nm, 1GHz Cortex A8 paired with a PowerVR SGX GPU tells me that Apple isn’t off its rocker. I don’t exactly know what Apple is doing with all of these CPU and GPU engineers in house, but licensing tech from the companies who have experience in building the architectures is still on the menu.
The A4 is a typical smartphone SoC. It’s got CPU, GPU and DRAM all on a single package. It also appears to be the same size as your typical smartphone SoC, meaning there’s a good chance we’ll see this thing in the next iPhone. The A4 has 256MB of on-package DRAM, just like the current iPhone 3GS SoC. Remember that memory is used for the OS, all applications as well as video memory. It’s a testament to just how lightweight the iPhone OS is.
Facebook
Twitter
RSS
108 Comments
View All Comments
TGressus - Thursday, April 8, 2010 - link
"I don't think anything anyone can write can convince you"That's a tough sell right there :( Nothing against your post.
This is the challenge all tablets have faced so far. That said, if anyone has the ability to succeed in this form factor it's the iCult.
Mike1111 - Thursday, April 8, 2010 - link
"While I realize that Atom hasn't been suited for such an application until now, there's no reason Apple should've picked the A4 over Moorestown. "I really think it was the right decision for Apple to go with ARM for the iPad, and that it's the right decision to stick with ARM for at least the next few years.
(1) Mobile version of Moorestown is not available yet, the netbook one draws too much power
(2) Apple has to use ARM in it's iPhone and iPod touch for the next few years, so for cross-device OS, app and SoC compatibility and development ARM was the right choice for the iPad (e.g. if Apple makes safari use the hardware better on the iPad, the iPhone and iPod touch will directly profit from it too)
(3) Moorestown would have been more expensive
(4) With ARM Apple can control and modify the CPU design as needed, they have total control. And Apple likes that.
(5a) There is a clear upgrade path for ARM with the Cortex-A9 and a multi-core version of it.
(5b) The A4 is already fast enough for most people and most iPad tasks (how many reviews mention that the iPad is unpleasantly slow, even for the average consumer?)
It's a real possibility that an iPad with a more optimized OS and safari, a better utilized (and more programmable) GPU (like the SGX545 with OpenCL etc.) and a dual-core Cortex-A9 @1.x GHz will improve the browsing performance beyond that of a netbook. And IF APPLE SAW THE NEED FOR IT (but I don't think they do), it could happen as early as next year (other ARM SoCs like Tegra2, OMAP4 and the dual-core Snapdragon will be available by then with comparable specs, so Apple should be able to pull it off too). Intel's Atom will still be a more power consuming, more expensive and way bigger multi-chip system early next year. And beyond that are ARM quad-core CPUs and dual-core GPUs...
It will take Atom at least 4 years to overtake ARM, in the areas that count for dedicated smartphones/slates/tablets, if at all. At some point, real life browsing will just be fast enough on an ARM slate and Intel netbook, so that 99% don't care about browsing speed as a feature anymore (like it will happen with video thanks to all these powerful but small dedicated decoders/encoders like IMG's VXD/VXE).
softdrinkviking - Thursday, April 8, 2010 - link
are you saying that there is no clear upgrade path for intel's moorstown line?and that apple has complete control over the future of ARM?
i didn't know they had already gone that far.
i seem to remember apple fumbling out of the power series of CPUs and crawling into into intel's arms when they realized it was the way to go.
there is nothing to prevent apple from switching to moorstown in a couple of generations, and they have proven that they are willing to make those kind of changes when the situation demands it.
Mike1111 - Friday, April 9, 2010 - link
"are you saying that there is no clear upgrade path for intel's moorstown line?"Of course not. I'm just saying that even if you think the ARM Cortex-A8 used in the iPad is not powerful enough, that there are clear CPU upgrades coming your way for years to come (higher frequencies, Cortex-A9, dual-core Cortex-A9, quad-core Cortex-A9). It's not like Apple is using a quad-core Cortex-A9 @ 4 GHz in 22nm right now, with no better ARM architecture on the horizon and standard lithography reaching a dead end...
"and that apple has complete control over the future of ARM? i didn't know they had already gone that far."
Sorry, what I meant was that Apple has complete control over how they implement the ARMv7 architecture in a chip (since Apple has most likely an architecture license like Qualcomm).
"there is nothing to prevent apple from switching to moorstown in a couple of generations, and they have proven that they are willing to make those kind of changes when the situation demands it."
I agree that Apple can and most likely will switch to Intel if their chips are clearly superior in all the ways that matter for a mobile product. I just think that's at least 4-5 years out for retail devices so there's no point in talking about how Apple should have used Moorestown for this year's iPad... or any iPad for that matter. I don't even think that Moorestown's successor will be ready (Medfield). Maybe Medfield's successor's successor in 22nm will be clearly superior (although parity could be reached a generation earlier). We'll see.
MadAd - Thursday, April 8, 2010 - link
Its huge. Way too big for car centre consoles. If they cut it in half, so its mid way between that and an Iphone then I would definitely want to upgrade my existing car PC.Its got all the right airs and graces to be a super satnav/speedcam/music/incar wifi unit, just need to cut the size and sell it with some kind of some kind of quick release device for power and cabling (or ill fabricate one) and its a winner.
teng029 - Thursday, April 8, 2010 - link
"Companies like Crestron and AMX supply ridiculously poor touch screen interfaces to their very expensive home automation installations."How exactly did you come to this conclusion? Have you extensively either used or program a control system touch panel?
Anand Lal Shimpi - Thursday, April 8, 2010 - link
Used, but not programmed. Those touchscreen controllers are just not in the same league in terms of UI as the iPad/iPhone honestly. I haven't used the latest incarnations but from the looks of them, they haven't changed tremendously.Take care,
Anand
athreya - Thursday, April 8, 2010 - link
1. Does writepad (phatware) or sketchbook pro (autodesk) allow one to take handwritten notes on the ipad? As in, im not looking for handwriting conversion to pages or word but can notes be taken and emailed across in the body of say a gmail or a Mail message? WHich stylus is the best for the ipad?2. How do you think iphone os 4.0 will solve the multitasking problem?
3. Between the wireless keyboard and the keyboard with dock which would you recommend and why? Will ANY BT keyboard work with the ipad?
4. Can you tell us how good it is at projecting powerpoint ppts onto a standard VGA projector? Does it support Office for Mac yet?
thanks a lot Anand. Terrific balanced review as always.
strikeback03 - Friday, April 9, 2010 - link
I doubt any stylus works with the iPad, due to the capacitive screen.dagamer34 - Friday, April 9, 2010 - link
A capacitive stylus would work.