You certainly would not use the card that came with the device for running an app since those default cards are usually like Class 2, not the Class 10 cards that Android fanatics boast about.
First of all, most of these cheap class 10 cards are slow. That's because being class 10 doesn't say anything about the flash inside. It's a topology for the bus. As we should all know, bus speed limits the maximum speeds, but doesn't guarantee them.
Many of these cheap cards are coming from China, from unknown manufacturers, or are even counterfeits, something that's a very big problem with flash cards. The biggest manufacturers such as Lexar and SanDisk have a major problem with this.
In fact, on Rob Galbraith's site, where they have been testing these cards for years, they would point out counterfeit cards. Even good camera stores, and places such as Staples, have had problems with counterfeit cards.
The point here is that these people want cheap cards. They aren't going to buy expensive cards, as that's the entire point of having a phone, or tablet with less memory (expensive), having the ability to utilize cards (cheap memory) for more storage. That way they can have a lower price product, and people buying that product can, mistakenly, think they are getting a bargain.
True in most cases the SD memory is considerably faster than the internal device memory which is designed for minimal power consumption rather than speed. Reading and writing to a large SD card will use up more battery even if it is faster.
Apple has been working on these processors for years as well. Only the first phone didn't have their stamp on it. Every phone since then has seen increasing Apple IP. It's not difficult to believe this. Apple now owns three proccessor companies.
And don't forget that Apple and Acorn, together, formed ARM specifically for the purpose of designing a mobile chip for a computer. That computer was the Newton, the first tablet, and the first device to use a mobile ARM chip. In fact, Apple owned half of ARM, until they gradually sold off the stock. Too bad about that too.
But Apple has had plenty of experience designing chips over the decades. They wrote much of the microcode for the PPC, and we're instrumental in combining the Power architecture with the Motorola architecture way back then.
Apple has a whale of a lot of knowledge about chip technology. Remember that they even designed their own chipsets for many years. And they bough a couple of GPU design companies.
They likely have at least as much knowledge as any other ARM manufacturer, and probably more than some, such as Nvidia, who came late to the game.
I agree that they are likely more knowledgable than nvidia, ti or sammy when it comes to arm based chips. Qualcomm is a little different, all they do is wireless and arm, i would put them a level above apple on the arm front and as far as wireless radios apple doesnt even try to compete with Q. Its nice to see that they apear to have a good working relationship.
I suspect that we might see the A6 used in a iPad Mini, come October... Possibly the next ATV, depending on what they do for games, etc.
I think that any new chip for the next iPad will require more RAM, more powerful GPU, and (possibly) more CPU cores.
If Apple has taken this direction in designing their own CPUs, conceivably, they could tailor the number of CPU cores on an A chip to suit the usage needs of a particular iDevice.
Another big advantage is that Apple could select different foundries to manufacture the chip -- and not be dependent on a competitor.
It's assumed that the A6 is a two core chip, but as Anand says, we can't be sure of that yet. When the x-rays are available, then we'll know for sure. But even before that, as soon as the phone is out, there will be testing on some sites that will let us know a bunch of things.
I doubt this is a four core chip with two cores disabled though.
It's always possible that we'll see an A6x for the new-er iPad.
True in most cases the SD memory is considerably faster than the internal device memory which is designed for minimal power consumption rather than speed. Reading and writing to a large SD card will use up more battery even if it is faster.
I agree that they are likely more knowledgable than nvidia, ti or sammy when it comes to arm based chips. Qualcomm is a little different, all they do is wireless and arm, i would put them a level above apple on the arm front and as far as wireless radios apple doesnt even try to compete with Q. Its nice to see that they apear to have a good working relationship.
Don't forget too, that Apple is a very large company, and can afford to have large chip teams. They even hired a major person from AMD last year.
As we saw, when "antenna gate" happened, that Apple has testing facilities for mobile antenna design that's been described as better than most antenna companies have. I wouldn't be surprised to find their facilities for SoC design is also well up there.
Amusingly, Apple could buy AMd with just 25% of a quarter's cash flow.
Don't forget too, that Apple is a very large company, and can afford to have large chip teams. They even hired a major person from AMD last year.
As we saw, when "antenna gate" happened, that Apple has testing facilities for mobile antenna design that's been described as better than most antenna companies have. I wouldn't be surprised to find their facilities for SoC design is also well up there.
Amusingly, Apple could buy AMd with just 25% of a quarter's cash flow.
With apple huge amount of cash the could buy just about any company they wanted. AMD would be an interesting purchase... Ive wondered about it before but I think they enjoy there preferential treatment from intel.
If so, I could see a symbiotic, mutually-beneficial relationship between Apple and IBM – where IBM manufactures chips for iPhones and iPads -- and IBM resells and supports these devices as part of its offerings/services to enterprise.
Aside... Do you ever wonder which 30 Fortune 500 companies are not testing or deploying iPads?
Which part?
1. SD is faster
2. Internal memory is design for low power usage
3. SD card uses more battery
4. All of the above?
That's a very simplistic way of looking at it. And it still isn't correct. You make an assumption that's wrong from the very beginning. That's assuming that all cards use a lot of battery power. That's not true. Secondly, you're assumption that internal flash is slow is also not true.
Most external SD cards, and there are a lot of different types these days, and a lot of devices won't even take a basic SD card, are very slow. They're rated in comparison to CD speeds. So a 1x card would be 150,000 KBs. A 4x card, which is what most cards that aren't labeled for actual speeds are, is just a miserly 600,000 KBs.
If we talk about medium fast cards, they're still not that fast. A 100x speed card is still just 15 MBs. But the interface to the device isn't likely to allow speeds above that, no matter what the card is rated for.
Don't forget too, that Apple is a very large company, and can afford to have large chip teams. They even hired a major person from AMD last year.
As we saw, when "antenna gate" happened, that Apple has testing facilities for mobile antenna design that's been described as better than most antenna companies have. I wouldn't be surprised to find their facilities for SoC design is also well up there.
Amusingly, Apple could buy AMd with just 25% of a quarter's cash flow.
With apple huge amount of cash the could buy just about any company they wanted. AMD would be an interesting purchase... Ive wondered about it before but I think they enjoy there preferential treatment from intel.
I remember [I think it was] when Apple went public... Steve Jobs joked that Apple was going to buy IBM and Xerox... Today, they could!
With apple huge amount of cash the could buy just about any company they wanted. AMD would be an interesting purchase... Ive wondered about it before but I think they enjoy there preferential treatment from intel.
Apple was burned first by Motorola around 2000, when they failed to deliver increased speeds for the G4 Apple was buying from them. That, by the way, was the beginning of the end for Motorola. Then, after Motorola sold off their chop division, Apple has serious problems in getting processors on time, and at speed. That's why they moved to IBM, as they couldn't get a fast enough dual core. But even IBM was chip constrained.
There was never a chance that Apple could move to AMD other than for a small proportion of their needs, because AMD has continual problems in delivering product on time, and at the performance levels promised.
I wonder how much the Anobit purchase affected the other ARM processors that can't use that technology anymore.
I'm under the impression Andobit was bought for their NAND prowess, specially in terms of reliability (hopefully for better reliability for faster and/or smaller NAND). I don't recall reading anything about them and ARM processors.
I wonder how much the Anobit purchase affected the other ARM processors that can't use that technology anymore.
I'm under the impression Andobit was bought for their NAND prowess, specially in terms of reliability (hopefully for better reliability for faster and/or smaller NAND). I don't recall reading anything about them and ARM processors.
I think he was referring to the fact that Apple and other ARM chips use Andobit NAND controllers in their SOCs.
From an Apple go to market strategy POV, it doesn't matter. Only the tech cognoscenti care. Apple never has tried to sell the iPhone on specs. That is why the Samsung ad in the story next to this misses the mark. People don't in general buy iOS devices for the specs, they buy it for the experience and the "it just works" ecosystem.
I am as technical as the next guy who's not actually a real hardware designer. (software architect/designer, strong OS background, good coder/developer, build my own VMware/Linux servers and deploy them in my home lab, etc.). I never used Apple devices until my gateway drug of a refurb 2nd gen iPod Touch from a woot.com impulse purchase a few years ago. (was a dedicated Palm guy before that). As soon as I started using it, I got it. Wow, this feels nice, was my immediate reaction. Made my old Palm TX feel just clunky. Then I took the next leap from Windows laptop to MacBook Pro Retina in June on a first day order. Now Windows feels old, clunky, and annoying.
I also have a pretty powerful android cell phone that my employer provided. That feels clunky and lame next to my iPhone 4s.
THAT is the experience I'm talking about and what Apple sells.
I think he was referring to the fact that Apple and other ARM chips use Andobit NAND controllers in their SOCs.
I'm not familiar with NAND being ARM SoCs or Apple A-chip ASICs.
Additionally, flash storage reliability drops quickly with long-term use. Single-level cell designs last up to several years, but the multi-level cell designs that have increased in popularity for mobile devices due to increased storage density actually decrease the useable life span of flash chips.
Anobit to the rescue
That's where Anobit and its technology come in. Anobit has developed unique technologies that can increase the reliability of multi-level cell designs. In fact, Apple already uses an Anobit-designed DSP chip in iPhones, iPads, and MacBook Airs to extend the life of the NAND flash chips in those devices.
Anobit—like Apple's other recent silicon design acquisitions, PA Semi and Intrinsity—is a fabless design house. Its specialty is creating, testing, and verifying new designs that implement its technological innovations, and then licensing the designs to companies like Apple. By buying up Anobit, Apple can keep its flash storage improvement technologies all to itself as a competitive advantage.
And yet they also like keeping secrets. This could be a secret worth keeping as long as they can -- it's a safe bet that if they did something good here, people will try to copy it.
Actually copying Apple here will be very difficult. It takes a lot of talent to design at this level. Further this has been a long time coming, even before Apple acquired PA Semi & Intrisity they had engineers working on CPU architecture and patenting concepts that could very well be implemented in this chip. Not to forget, we need to mention IP from the Anobit purchase.
Further this is the iPhone, which is probably an implementation optimized for low power and performance. This could be a very interesting upgrade to the next iPad, where everything from Flash access to the GPU is much faster. The thought of a 2X or more faster iPad next year is in fact more interesting to me than the iPhone at this point.
Which probably brings up a point, Apple will likely try to keep the details quite with this SoC until after the iPad variant is released next year. I just have this feeling there is a lot of customization in this chip.
It should be pointed out that we do not have any proof at all as to what is in A6. However you are extremely negative on Apple here, they have been working on this for years. Just follow the patent record.
Unless they paid Qualcomm for IP or subcontracted some of the work to Q I doubt apple made a Krait like custom core.
You totally misunderstand what is being said here. The idea is that Apple took an approach that was like Qualcomms in that they did a custom core, they did not use Qualcomms core.
It took qualcomms army of engineers years to develop the first snapdragon processor, they're first custom SOC with a custom core. As much as I believe in apples innovative abilities I just doubt they built a krait like processor of there first try at a custom core.
It wouldn't make sense otherwise. You don't go around buying up specialized companies to build me too hardware.
Also, if they did build a krait type cpu then why didnt they incorporate the radio into the chip like qualcomm did? That would have been very apple-like to put it all in the same piece of silicon and they're already buying the chip from qualcomm so why not license it from them and just build it into the SoC.
I'm not sure what you're trying to show me in that quoted text. I see nothing about ARM.SoCs or Apple ASICs.
NAND for the primary storage tends to quite large. The NAND in the iPhone 4S is 24nm and it's still not small. It might be bigger than the A5 chip.
iFixit oddly didn't have a good pic of the back side of the iPhone 5 logic board so here is the board from from the iPhone 4. The A4 is obvious and the NAND is that big ol' chip on the reverse side of the board. It's close to the A4 but it's not part of the ASIC.
Comments
First of all, most of these cheap class 10 cards are slow. That's because being class 10 doesn't say anything about the flash inside. It's a topology for the bus. As we should all know, bus speed limits the maximum speeds, but doesn't guarantee them.
Many of these cheap cards are coming from China, from unknown manufacturers, or are even counterfeits, something that's a very big problem with flash cards. The biggest manufacturers such as Lexar and SanDisk have a major problem with this.
In fact, on Rob Galbraith's site, where they have been testing these cards for years, they would point out counterfeit cards. Even good camera stores, and places such as Staples, have had problems with counterfeit cards.
The point here is that these people want cheap cards. They aren't going to buy expensive cards, as that's the entire point of having a phone, or tablet with less memory (expensive), having the ability to utilize cards (cheap memory) for more storage. That way they can have a lower price product, and people buying that product can, mistakenly, think they are getting a bargain.
That's totally false.
Quote:
Originally Posted by melgross
Apple has been working on these processors for years as well. Only the first phone didn't have their stamp on it. Every phone since then has seen increasing Apple IP. It's not difficult to believe this. Apple now owns three proccessor companies.
And don't forget that Apple and Acorn, together, formed ARM specifically for the purpose of designing a mobile chip for a computer. That computer was the Newton, the first tablet, and the first device to use a mobile ARM chip. In fact, Apple owned half of ARM, until they gradually sold off the stock. Too bad about that too.
But Apple has had plenty of experience designing chips over the decades. They wrote much of the microcode for the PPC, and we're instrumental in combining the Power architecture with the Motorola architecture way back then.
Apple has a whale of a lot of knowledge about chip technology. Remember that they even designed their own chipsets for many years. And they bough a couple of GPU design companies.
They likely have at least as much knowledge as any other ARM manufacturer, and probably more than some, such as Nvidia, who came late to the game.
I agree that they are likely more knowledgable than nvidia, ti or sammy when it comes to arm based chips. Qualcomm is a little different, all they do is wireless and arm, i would put them a level above apple on the arm front and as far as wireless radios apple doesnt even try to compete with Q. Its nice to see that they apear to have a good working relationship.
It's assumed that the A6 is a two core chip, but as Anand says, we can't be sure of that yet. When the x-rays are available, then we'll know for sure. But even before that, as soon as the phone is out, there will be testing on some sites that will let us know a bunch of things.
I doubt this is a four core chip with two cores disabled though.
It's always possible that we'll see an A6x for the new-er iPad.
Quote:
Originally Posted by melgross
Quote:
Originally Posted by mstone
True in most cases the SD memory is considerably faster than the internal device memory which is designed for minimal power consumption rather than speed. Reading and writing to a large SD card will use up more battery even if it is faster.
That's totally false.
Which part?
1. SD is faster
2. Internal memory is design for low power usage
3. SD card uses more battery
4. All of the above?
Don't forget too, that Apple is a very large company, and can afford to have large chip teams. They even hired a major person from AMD last year.
As we saw, when "antenna gate" happened, that Apple has testing facilities for mobile antenna design that's been described as better than most antenna companies have. I wouldn't be surprised to find their facilities for SoC design is also well up there.
Amusingly, Apple could buy AMd with just 25% of a quarter's cash flow.
Quote:
Originally Posted by melgross
Don't forget too, that Apple is a very large company, and can afford to have large chip teams. They even hired a major person from AMD last year.
As we saw, when "antenna gate" happened, that Apple has testing facilities for mobile antenna design that's been described as better than most antenna companies have. I wouldn't be surprised to find their facilities for SoC design is also well up there.
Amusingly, Apple could buy AMd with just 25% of a quarter's cash flow.
With apple huge amount of cash the could buy just about any company they wanted. AMD would be an interesting purchase... Ive wondered about it before but I think they enjoy there preferential treatment from intel.
Kinda' OT…
Does IBM still have its own foundries?
If so, I could see a symbiotic, mutually-beneficial relationship between Apple and IBM – where IBM manufactures chips for iPhones and iPads -- and IBM resells and supports these devices as part of its offerings/services to enterprise.
Aside... Do you ever wonder which 30 Fortune 500 companies are not testing or deploying iPads?
Dictated on my iMac
That's a very simplistic way of looking at it. And it still isn't correct. You make an assumption that's wrong from the very beginning. That's assuming that all cards use a lot of battery power. That's not true. Secondly, you're assumption that internal flash is slow is also not true.
Most external SD cards, and there are a lot of different types these days, and a lot of devices won't even take a basic SD card, are very slow. They're rated in comparison to CD speeds. So a 1x card would be 150,000 KBs. A 4x card, which is what most cards that aren't labeled for actual speeds are, is just a miserly 600,000 KBs.
If we talk about medium fast cards, they're still not that fast. A 100x speed card is still just 15 MBs. But the interface to the device isn't likely to allow speeds above that, no matter what the card is rated for.
Quote:
Originally Posted by WESALLEN
Quote:
Originally Posted by melgross
Don't forget too, that Apple is a very large company, and can afford to have large chip teams. They even hired a major person from AMD last year.
As we saw, when "antenna gate" happened, that Apple has testing facilities for mobile antenna design that's been described as better than most antenna companies have. I wouldn't be surprised to find their facilities for SoC design is also well up there.
Amusingly, Apple could buy AMd with just 25% of a quarter's cash flow.
With apple huge amount of cash the could buy just about any company they wanted. AMD would be an interesting purchase... Ive wondered about it before but I think they enjoy there preferential treatment from intel.
I remember [I think it was] when Apple went public... Steve Jobs joked that Apple was going to buy IBM and Xerox... Today, they could!
Apple was burned first by Motorola around 2000, when they failed to deliver increased speeds for the G4 Apple was buying from them. That, by the way, was the beginning of the end for Motorola. Then, after Motorola sold off their chop division, Apple has serious problems in getting processors on time, and at speed. That's why they moved to IBM, as they couldn't get a fast enough dual core. But even IBM was chip constrained.
There was never a chance that Apple could move to AMD other than for a small proportion of their needs, because AMD has continual problems in delivering product on time, and at the performance levels promised.
I'm under the impression Andobit was bought for their NAND prowess, specially in terms of reliability (hopefully for better reliability for faster and/or smaller NAND). I don't recall reading anything about them and ARM processors.
I think he was referring to the fact that Apple and other ARM chips use Andobit NAND controllers in their SOCs.
I am as technical as the next guy who's not actually a real hardware designer. (software architect/designer, strong OS background, good coder/developer, build my own VMware/Linux servers and deploy them in my home lab, etc.). I never used Apple devices until my gateway drug of a refurb 2nd gen iPod Touch from a woot.com impulse purchase a few years ago. (was a dedicated Palm guy before that). As soon as I started using it, I got it. Wow, this feels nice, was my immediate reaction. Made my old Palm TX feel just clunky. Then I took the next leap from Windows laptop to MacBook Pro Retina in June on a first day order. Now Windows feels old, clunky, and annoying.
I also have a pretty powerful android cell phone that my employer provided. That feels clunky and lame next to my iPhone 4s.
THAT is the experience I'm talking about and what Apple sells.
I'm not familiar with NAND being ARM SoCs or Apple A-chip ASICs.
http://arstechnica.com/apple/2011/12/apple-lays-down-half-a-billion-to-secure-its-flash-storage-future/
Actually copying Apple here will be very difficult. It takes a lot of talent to design at this level. Further this has been a long time coming, even before Apple acquired PA Semi & Intrisity they had engineers working on CPU architecture and patenting concepts that could very well be implemented in this chip. Not to forget, we need to mention IP from the Anobit purchase.
Further this is the iPhone, which is probably an implementation optimized for low power and performance. This could be a very interesting upgrade to the next iPad, where everything from Flash access to the GPU is much faster. The thought of a 2X or more faster iPad next year is in fact more interesting to me than the iPhone at this point.
Which probably brings up a point, Apple will likely try to keep the details quite with this SoC until after the iPad variant is released next year. I just have this feeling there is a lot of customization in this chip.
You totally misunderstand what is being said here. The idea is that Apple took an approach that was like Qualcomms in that they did a custom core, they did not use Qualcomms core. It wouldn't make sense otherwise. You don't go around buying up specialized companies to build me too hardware. Flexibility?
I'm not sure what you're trying to show me in that quoted text. I see nothing about ARM.SoCs or Apple ASICs.
NAND for the primary storage tends to quite large. The NAND in the iPhone 4S is 24nm and it's still not small. It might be bigger than the A5 chip.
iFixit oddly didn't have a good pic of the back side of the iPhone 5 logic board so here is the board from from the iPhone 4. The A4 is obvious and the NAND is that big ol' chip on the reverse side of the board. It's close to the A4 but it's not part of the ASIC.