Intel is not exactly taking the market by storm with their smaller process technology. Yet. Ivy Bridge has been delayed. Their lower power Atom processors on 32nm are nothing special. I doubt they could match the A5X in power efficiency when integrated into a tablet. There is a lot more that goes into efficient chip design than just shrinking the process.
Shrink the process is how Intel keeps ahead of the game. It isn't like their processor designs are in anyway outstanding.
Quote:
I see the A5X as being similar to the "tick tock" strategy Intel has been using - first you get the design right, then you shrink it. Apple has the ability to create processors that include unique feature sets. Beside the 2 CPU & 4 GPU cores, it has circuity for interfacing with the cameras and memory.
If the processor is still on 45nm then I'm a bit shocked. However my iPad 3 experience has been really good so I really don't care. Lower power and more speed is always nice but I'm impressed with the step up from iPad 1.
Quote:
What is interesting about A5X is they moved the DRAM off to separate chips, which would seem to indicate it would not work in an iPhone because of the extra space required. I could see Apple have two distinct lines of ARM processors in the future - ones focused for iPhones with RAM onboard, and the other on iPads with separate RAM chips. It is probably cheaper to manufacturer a chip without piggybacking the RAM on another layer, and it gives Apple flexibility to source iPad RAM from multiple vendors
I would think that a dual track development program is a requirement. it is far more important to stuff tablets with more power. More importantly I think Apple needs an even smaller processor for phones and other smaller devices.
There are no infrared microscope pictures here. The first picture is made on an optical microscope and the second one is made on a SEM (scanning electron microscope).
I wouldn't call it an optical microscope, someone might have just used an iPhone and then zoomed in/ cropped in Photoshop.
Back in my day, optical microscopes were... yada yada yada
Quote:
Originally Posted by msimpson
Intel is not exactly taking the market by storm with their smaller process technology. Yet. Ivy Bridge has been delayed. Their lower power Atom processors on 32nm are nothing special. I doubt they could match the A5X in power efficiency when integrated into a tablet. There is a lot more that goes into efficient chip design than just shrinking the process.
I see the A5X as being similar to the "tick tock" strategy Intel has been using - first you get the design right, then you shrink it. Apple has the ability to create processors that include unique feature sets. Beside the 2 CPU & 4 GPU cores, it has circuity for interfacing with the cameras and memory.
What is interesting about A5X is they moved the DRAM off to separate chips, which would seem to indicate it would not work in an iPhone because of the extra space required. I could see Apple have two distinct lines of ARM processors in the future - ones focused for iPhones with RAM onboard, and the other on iPads with separate RAM chips. It is probably cheaper to manufacturer a chip without piggybacking the RAM on another layer, and it gives Apple flexibility to source iPad RAM from multiple vendors
The chilling possibility is that short of Apple making its own fab, any Apple A6/7/etc. at 28nm or less may very well have to be done through Intel... Unless Samsung or others can really deliver ~ we'll have to see, I'm not saying Samsung sucks per se simply because this is at 45nm. But one would imagine a quad PowerVR Rogue with ARM quad (big.Little dual A7 with dual A15) would do much better on a nice 28nm process.
BTW besides space, did they separate it also because they need to have a heatsink on the main die, and no need for heatsinks on the RAM?
Still on 45nm? Why does Apple still let Samsung fab them? Can't Apple design them but have Intel fabricate them? Intel had 32nm since 2010 and they are bringing out 22nm this year.
Because the only third parties that Intel lets use its fabs are FPGA companies, and that's because FPGAs are simpler devices that allow Intel to test their new processes earlier than they can with more complex CPU designs (i.e., there's a real benefit for Intel besides earning money for letting FPGAs be fabbed on their leading-edge processes).
In addition you can't just switch to a different fab. Each company's process is different from the next. It can take 6 months to a year to port a design to a different fab.
And ultimately, Apple needs millions of these chips each month, and thus proven processes like Samsung's 45nm process are guaranteed to allow that rate of production with no hiccups (e.g., TSMC 28nm issues).
Samsung do currently have a 32nm process as well, so I'm sure the A6 will be using that once it is ready to handle the volumes that Apple require. I'm sure that Apple is also looking at GlobalFoundries and TSMC as alternatives as well.
Samsung Electronic's market cap is something under $180 B. Typically, acquisitions are made at a slight premium to market cap, so it would take Apple something like $200 B to buy them.
With $100 B in cash and no debt, Apple would have absolutely no problem in buying Samsung if they wished. Of course, it is extremely unlikely that they'd want to.
Samsung is not for sale due to Korean government regulations, you can't even buy Samsung stock in US on open market, but only though specialized brokers who have access to Korean stock market.
But Intel is available. Its market cap is $138 B. Also AMD, TI, Transmeta, National Semi, Taiwan Semi.
AMD can build CPU and flash memory and it's only $6 B. But of course it will take a couple of years until any of these would be able to produce processors in these high volumes.
Samsung Electronic's market cap is something under $180 B. Typically, acquisitions are made at a slight premium to market cap, so it would take Apple something like $200 B to buy them.
With $100 B in cash and no debt, Apple would have absolutely no problem in buying Samsung if they wished. Of course, it is extremely unlikely that they'd want to.
Buying another company for the mear fact that it can and based on personal grudge isnt a very sound investment move. If you were CEO, you'd lose your job in an instant.
Comments
Intel is not exactly taking the market by storm with their smaller process technology. Yet. Ivy Bridge has been delayed. Their lower power Atom processors on 32nm are nothing special. I doubt they could match the A5X in power efficiency when integrated into a tablet. There is a lot more that goes into efficient chip design than just shrinking the process.
Shrink the process is how Intel keeps ahead of the game. It isn't like their processor designs are in anyway outstanding.
I see the A5X as being similar to the "tick tock" strategy Intel has been using - first you get the design right, then you shrink it. Apple has the ability to create processors that include unique feature sets. Beside the 2 CPU & 4 GPU cores, it has circuity for interfacing with the cameras and memory.
If the processor is still on 45nm then I'm a bit shocked. However my iPad 3 experience has been really good so I really don't care. Lower power and more speed is always nice but I'm impressed with the step up from iPad 1.
What is interesting about A5X is they moved the DRAM off to separate chips, which would seem to indicate it would not work in an iPhone because of the extra space required. I could see Apple have two distinct lines of ARM processors in the future - ones focused for iPhones with RAM onboard, and the other on iPads with separate RAM chips. It is probably cheaper to manufacturer a chip without piggybacking the RAM on another layer, and it gives Apple flexibility to source iPad RAM from multiple vendors
I would think that a dual track development program is a requirement. it is far more important to stuff tablets with more power. More importantly I think Apple needs an even smaller processor for phones and other smaller devices.
45nm process
Other products are being manufactured at 28nm and 22nm.
So, how more do we get when the iPad/iPhone processors are made with these narrow widths?
Exponential progress really hurts the brain.
Well, 28nm and 22nm, if you're Intel, doesn't hurt 'em one bit. Samsung, TSMC, AMD, etc... not sure.
10nm or bust!
45nm explains some heat issues.
Dear editor,
There are no infrared microscope pictures here. The first picture is made on an optical microscope and the second one is made on a SEM (scanning electron microscope).
I wouldn't call it an optical microscope, someone might have just used an iPhone and then zoomed in/ cropped in Photoshop.
Back in my day, optical microscopes were... yada yada yada
Intel is not exactly taking the market by storm with their smaller process technology. Yet. Ivy Bridge has been delayed. Their lower power Atom processors on 32nm are nothing special. I doubt they could match the A5X in power efficiency when integrated into a tablet. There is a lot more that goes into efficient chip design than just shrinking the process.
I see the A5X as being similar to the "tick tock" strategy Intel has been using - first you get the design right, then you shrink it. Apple has the ability to create processors that include unique feature sets. Beside the 2 CPU & 4 GPU cores, it has circuity for interfacing with the cameras and memory.
What is interesting about A5X is they moved the DRAM off to separate chips, which would seem to indicate it would not work in an iPhone because of the extra space required. I could see Apple have two distinct lines of ARM processors in the future - ones focused for iPhones with RAM onboard, and the other on iPads with separate RAM chips. It is probably cheaper to manufacturer a chip without piggybacking the RAM on another layer, and it gives Apple flexibility to source iPad RAM from multiple vendors
The chilling possibility is that short of Apple making its own fab, any Apple A6/7/etc. at 28nm or less may very well have to be done through Intel... Unless Samsung or others can really deliver ~ we'll have to see, I'm not saying Samsung sucks per se simply because this is at 45nm. But one would imagine a quad PowerVR Rogue with ARM quad (big.Little dual A7 with dual A15) would do much better on a nice 28nm process.
BTW besides space, did they separate it also because they need to have a heatsink on the main die, and no need for heatsinks on the RAM?
Still on 45nm? Why does Apple still let Samsung fab them? Can't Apple design them but have Intel fabricate them? Intel had 32nm since 2010 and they are bringing out 22nm this year.
Because the only third parties that Intel lets use its fabs are FPGA companies, and that's because FPGAs are simpler devices that allow Intel to test their new processes earlier than they can with more complex CPU designs (i.e., there's a real benefit for Intel besides earning money for letting FPGAs be fabbed on their leading-edge processes).
In addition you can't just switch to a different fab. Each company's process is different from the next. It can take 6 months to a year to port a design to a different fab.
And ultimately, Apple needs millions of these chips each month, and thus proven processes like Samsung's 45nm process are guaranteed to allow that rate of production with no hiccups (e.g., TSMC 28nm issues).
Samsung do currently have a 32nm process as well, so I'm sure the A6 will be using that once it is ready to handle the volumes that Apple require. I'm sure that Apple is also looking at GlobalFoundries and TSMC as alternatives as well.
So what?
Samsung Electronic's market cap is something under $180 B. Typically, acquisitions are made at a slight premium to market cap, so it would take Apple something like $200 B to buy them.
With $100 B in cash and no debt, Apple would have absolutely no problem in buying Samsung if they wished. Of course, it is extremely unlikely that they'd want to.
Samsung is not for sale due to Korean government regulations, you can't even buy Samsung stock in US on open market, but only though specialized brokers who have access to Korean stock market.
But Intel is available. Its market cap is $138 B. Also AMD, TI, Transmeta, National Semi, Taiwan Semi.
AMD can build CPU and flash memory and it's only $6 B. But of course it will take a couple of years until any of these would be able to produce processors in these high volumes.
So what?
Samsung Electronic's market cap is something under $180 B. Typically, acquisitions are made at a slight premium to market cap, so it would take Apple something like $200 B to buy them.
With $100 B in cash and no debt, Apple would have absolutely no problem in buying Samsung if they wished. Of course, it is extremely unlikely that they'd want to.
Buying another company for the mear fact that it can and based on personal grudge isnt a very sound investment move. If you were CEO, you'd lose your job in an instant.