it is a stretch to say A6 will be based on ARMs A15 core. Apple could just as easily shrink the current design and add a couple of more cores. At 28nm they could go to four cores, up the clock rate and still be ahead on power.
Beyond that A15 IS NOT 64 BIT. Apple will need the 64 bit address space sooner than many think.
Oddly, your comments regarding what is hard to follow, are unclear. Making some assumptions, on what arguments you find unclear, here are follow up comments.
Given the progression between Apple's mobile core designs, I think it's reasonable and generally speculated that the A6 would be based upon the ARM A15. The mobile SOCs seem to be a step or two behind in terms of fabrication density, so a move to A15 is more likely than a move to 28nm. The current speculation is that Apple would push to 40nm with a chance of 28nm versus the current 45nm. I think it's far more speculative to say that a change in core count + change in fabrication by 2 steps + change in clock rate > A15 + change in fabrication by 1 step with respect to power and speed. A quick look at the history of ARM cores shows a marked improvement in performance/watt with each subsequent generation. The move from A8 to A9 resulted in a DMIPS/MHz improvement of 2 to 2.5 or 25%. There was a similar improvement moving from the A5 to the A8, so it seems reasonable to anticipate a similar improvement moving from the A9 to A15. Given relatively low operating frequencies of mobile chips, and the direct relationship between frequency and power, it seems more sensible to move to a more efficient core.
Apple to move ARM SoC production away from Samsung in 2012
With respect to the requirement for 64-bit address space, that requirement is only now beginning to exist on the desktop and even then, only in limited areas. When system memory on the iPhone/iPad/iPod is still <1GB and hasn't changed in 2 generations of product, it seems a bit much to suggest that 64-bit is much of a priority.
As for your final comment regarding "some debate going on about this", it's not clear which statement(s) you are making referencing.
Oddly, your comments regarding what is hard to follow, are unclear. Making some assumptions, on what arguments you find unclear, here are follow up comments.
it is you that I find unclear and hard to follow.
Quote:
Given the progression between Apple's mobile core designs, I think it's reasonable and generally speculated that the A6 would be based upon the ARM A15.
I see that as a stretch. No one is speculating about A15 going into Apples next chip. A15 is targeted at the server market.
Quote:
The mobile SOCs seem to be a step or two behind in terms of fabrication density, so a move to A15 is more likely than a move to 28nm.
not really, many 28nm Cortex A9 solutions have already taped out at 28nm. Beyond that Apple should be highly motivated to get to 28nm as soon as possible to lower power usage. In any event we are talking production for 2012.
Now I admit that this is aggressive and that Apple could stop at 32nm instead. The thing is they have to be aggressive process wise so that they can both boost performance and manage power.
Quote:
The current speculation is that Apple would push to 40nm with a chance of 28nm versus the current 45nm.
40nm? The next node is 32nm.
Quote:
I think it's far more speculative to say that a change in core count + change in fabrication by 2 steps + change in clock rate > A15 + change in fabrication by 1 step with respect to power and speed.
A15 seems to be loaded with functionality Apple doesn't need, staying with the A9 core allows them to maintain a trim hardware profile. A9 can go to four core easily with the process shrinks, keeping watts expended low.
Quote:
A quick look at the history of ARM cores shows a marked improvement in performance/watt with each subsequent generation. The move from A8 to A9 resulted in a DMIPS/MHz improvement of 2 to 2.5 or 25%. There was a similar improvement moving from the A5 to the A8, so it seems reasonable to anticipate a similar improvement moving from the A9 to A15. Given relatively low operating frequencies of mobile chips, and the direct relationship between frequency and power, it seems more sensible to move to a more efficient core.
So you believe the A15 core is more efficient? I'm not convinced though info is thin right now.
Quote:
Apple to move ARM SoC production away from Samsung in 2012
With respect to the requirement for 64-bit address space, that requirement is only now beginning to exist on the desktop and even then, only in limited areas. When system memory on the iPhone/iPad/iPod is still <1GB and hasn't changed in 2 generations of product, it seems a bit much to suggest that 64-bit is much of a priority.
My MBP from early 2008 is a 64 bit platform. Before that I was running 64 bit Linux for some time. I think you are out of touch. As to the less than 1GB of RAM in iOS devices that isn't exactly somthing to celebrate.
Quote:
As for your final comment regarding "some debate going on about this", it's not clear which statement(s) you are making referencing.
In any event I'm pretty much convinced that A6 will be Apples attempt to roll in a lot of their own iP. We should see TB built into the chip for example.
The A15 is targeted at everything from phones to servers. Not only its cores are 25-40% faster than A9 cores at the same speed, they will still be very efficient. The A15 will be available from 1C@1GHz to [email protected], multiple processors can be linked together (AMBA® 4 Cache Coherent Interconnect), and it will have (LPAE) extensions for up to 1TB of RAM support (one of the main advantage of 64-bit architecture). It won't compete in raw performance with 2012 Core ix cpus, but it will still be king in terms of power consumption. Don't know for sure what will Apple do with it or when, but expect huge performance improvements on current designs and maybe good enough performance for new ones.
The A15 is targeted at everything from phones to servers.
This right here has me wondering about the A15s success. I would imagine that a cell phone targeted version would vary drastically from the server variant. After all what is them point of virtualization in a cell phone?
Virtualization along with other features are why I think Apple might choose a different route. After all they are designing processors in order to minimize waste with unneeded functionality. Sure there are other reasons but the ability to tailor a processor to your exact needs is extremely valuable to Apple.
Quote:
Not only its cores are 25-40% faster than A9 cores at the same speed, they will still be very efficient. The A15 will be available from 1C@1GHz to [email protected], multiple processors can be linked together (AMBA® 4 Cache Coherent Interconnect), and it will have (LPAE) extensions for up to 1TB of RAM support (one of the main advantage of 64-bit architecture).
Frankly your statement about LPAE is bull. If that was the case we would all be running 32 bit systems with extended addressing. Those systems died off as soon 64 bit os'es became viable. 1TB. of RAM is pretty useless if processes can't address it.
Quote:
It won't compete in raw performance with 2012 Core ix cpus, but it will still be king in terms of power consumption. Don't know for sure what will Apple do with it or when, but expect huge performance improvements on current designs and maybe good enough performance for new ones.
I'd rather wait and see. A15 just doesn't strike me as the best processor to evolve IOS on. However ARM hardware is highly customizable and Apple has some really good engineers so maybe they can implement a good solution.
It will be interesting to see how this all unfolds.
...
It really doesn't matter who is right here...
What's the matter with you???
I was just describing a few features of the A15 design and providing a link of your "education" as you were mistaken in thinking that A15 was just targeted at servers.
Relax. Don't forget you're over 50, your heart is not as strong as it was.
You buy a Mac, dont come whinning when you cant play your favorite games on it! You better read up on what Macs can run!
Thanks for all the spam, by the way. It's not like we'll ever actually go to your links, given that you're spamming our website. How freaking stupid can you GET?
The A15 is targeted at everything from phones to servers. Not only its cores are 25-40% faster than A9 cores at the same speed, they will still be very efficient. The A15 will be available from 1C@1GHz to [email protected], multiple processors can be linked together (AMBA® 4 Cache Coherent Interconnect), and it will have (LPAE) extensions for up to 1TB of RAM support (one of the main advantage of 64-bit architecture). It won't compete in raw performance with 2012 Core ix cpus, but it will still be king in terms of power consumption. Don't know for sure what will Apple do with it or when, but expect huge performance improvements on current designs and maybe good enough performance for new ones.
Comments
it is a stretch to say A6 will be based on ARMs A15 core. Apple could just as easily shrink the current design and add a couple of more cores. At 28nm they could go to four cores, up the clock rate and still be ahead on power.
Beyond that A15 IS NOT 64 BIT. Apple will need the 64 bit address space sooner than many think.
There is some debate going on about this.
@wizard69
Oddly, your comments regarding what is hard to follow, are unclear. Making some assumptions, on what arguments you find unclear, here are follow up comments.
Given the progression between Apple's mobile core designs, I think it's reasonable and generally speculated that the A6 would be based upon the ARM A15. The mobile SOCs seem to be a step or two behind in terms of fabrication density, so a move to A15 is more likely than a move to 28nm. The current speculation is that Apple would push to 40nm with a chance of 28nm versus the current 45nm. I think it's far more speculative to say that a change in core count + change in fabrication by 2 steps + change in clock rate > A15 + change in fabrication by 1 step with respect to power and speed. A quick look at the history of ARM cores shows a marked improvement in performance/watt with each subsequent generation. The move from A8 to A9 resulted in a DMIPS/MHz improvement of 2 to 2.5 or 25%. There was a similar improvement moving from the A5 to the A8, so it seems reasonable to anticipate a similar improvement moving from the A9 to A15. Given relatively low operating frequencies of mobile chips, and the direct relationship between frequency and power, it seems more sensible to move to a more efficient core.
Apple to move ARM SoC production away from Samsung in 2012
http://bit.ly/kqV8Iv
ARM Cortex-A Series
http://bit.ly/mtonIJ
With respect to the requirement for 64-bit address space, that requirement is only now beginning to exist on the desktop and even then, only in limited areas. When system memory on the iPhone/iPad/iPod is still <1GB and hasn't changed in 2 generations of product, it seems a bit much to suggest that 64-bit is much of a priority.
As for your final comment regarding "some debate going on about this", it's not clear which statement(s) you are making referencing.
@wizard69
Oddly, your comments regarding what is hard to follow, are unclear. Making some assumptions, on what arguments you find unclear, here are follow up comments.
it is you that I find unclear and hard to follow.
Given the progression between Apple's mobile core designs, I think it's reasonable and generally speculated that the A6 would be based upon the ARM A15.
I see that as a stretch. No one is speculating about A15 going into Apples next chip. A15 is targeted at the server market.
The mobile SOCs seem to be a step or two behind in terms of fabrication density, so a move to A15 is more likely than a move to 28nm.
not really, many 28nm Cortex A9 solutions have already taped out at 28nm. Beyond that Apple should be highly motivated to get to 28nm as soon as possible to lower power usage. In any event we are talking production for 2012.
Now I admit that this is aggressive and that Apple could stop at 32nm instead. The thing is they have to be aggressive process wise so that they can both boost performance and manage power.
The current speculation is that Apple would push to 40nm with a chance of 28nm versus the current 45nm.
40nm? The next node is 32nm.
I think it's far more speculative to say that a change in core count + change in fabrication by 2 steps + change in clock rate > A15 + change in fabrication by 1 step with respect to power and speed.
A15 seems to be loaded with functionality Apple doesn't need, staying with the A9 core allows them to maintain a trim hardware profile. A9 can go to four core easily with the process shrinks, keeping watts expended low.
A quick look at the history of ARM cores shows a marked improvement in performance/watt with each subsequent generation. The move from A8 to A9 resulted in a DMIPS/MHz improvement of 2 to 2.5 or 25%. There was a similar improvement moving from the A5 to the A8, so it seems reasonable to anticipate a similar improvement moving from the A9 to A15. Given relatively low operating frequencies of mobile chips, and the direct relationship between frequency and power, it seems more sensible to move to a more efficient core.
So you believe the A15 core is more efficient? I'm not convinced though info is thin right now.
Apple to move ARM SoC production away from Samsung in 2012
http://bit.ly/kqV8Iv
ARM Cortex-A Series
http://bit.ly/mtonIJ
With respect to the requirement for 64-bit address space, that requirement is only now beginning to exist on the desktop and even then, only in limited areas. When system memory on the iPhone/iPad/iPod is still <1GB and hasn't changed in 2 generations of product, it seems a bit much to suggest that 64-bit is much of a priority.
My MBP from early 2008 is a 64 bit platform. Before that I was running 64 bit Linux for some time. I think you are out of touch. As to the less than 1GB of RAM in iOS devices that isn't exactly somthing to celebrate.
As for your final comment regarding "some debate going on about this", it's not clear which statement(s) you are making referencing.
In any event I'm pretty much convinced that A6 will be Apples attempt to roll in a lot of their own iP. We should see TB built into the chip for example.
Um... I was being facetious with the entire post. Just in case the obviousness of my predictions wasn't so obvious.
Sometimes the forums can be pretty tone deaf, leading to slightly surreal results.
A15 is targeted at the server market.
The A15 is targeted at everything from phones to servers. Not only its cores are 25-40% faster than A9 cores at the same speed, they will still be very efficient. The A15 will be available from 1C@1GHz to [email protected], multiple processors can be linked together (AMBA® 4 Cache Coherent Interconnect), and it will have (LPAE) extensions for up to 1TB of RAM support (one of the main advantage of 64-bit architecture). It won't compete in raw performance with 2012 Core ix cpus, but it will still be king in terms of power consumption. Don't know for sure what will Apple do with it or when, but expect huge performance improvements on current designs and maybe good enough performance for new ones.
http://www.arm.com/products/processo...cortex-a15.php
The A15 is targeted at everything from phones to servers.
This right here has me wondering about the A15s success. I would imagine that a cell phone targeted version would vary drastically from the server variant. After all what is them point of virtualization in a cell phone?
Virtualization along with other features are why I think Apple might choose a different route. After all they are designing processors in order to minimize waste with unneeded functionality. Sure there are other reasons but the ability to tailor a processor to your exact needs is extremely valuable to Apple.
Not only its cores are 25-40% faster than A9 cores at the same speed, they will still be very efficient. The A15 will be available from 1C@1GHz to [email protected], multiple processors can be linked together (AMBA® 4 Cache Coherent Interconnect), and it will have (LPAE) extensions for up to 1TB of RAM support (one of the main advantage of 64-bit architecture).
Frankly your statement about LPAE is bull. If that was the case we would all be running 32 bit systems with extended addressing. Those systems died off as soon 64 bit os'es became viable. 1TB. of RAM is pretty useless if processes can't address it.
It won't compete in raw performance with 2012 Core ix cpus, but it will still be king in terms of power consumption. Don't know for sure what will Apple do with it or when, but expect huge performance improvements on current designs and maybe good enough performance for new ones.
I'd rather wait and see. A15 just doesn't strike me as the best processor to evolve IOS on. However ARM hardware is highly customizable and Apple has some really good engineers so maybe they can implement a good solution.
http://www.arm.com/products/processo...cortex-a15.php
It really doesn't matter who is right here as the move to 28nm ought to allow for one very interesting iPad3.
It will be interesting to see how this all unfolds.
...
It really doesn't matter who is right here...
What's the matter with you???
I was just describing a few features of the A15 design and providing a link of your "education" as you were mistaken in thinking that A15 was just targeted at servers.
Relax. Don't forget you're over 50, your heart is not as strong as it was.
You buy a Mac, dont come whinning when you cant play your favorite games on it! You better read up on what Macs can run!
Thanks for all the spam, by the way. It's not like we'll ever actually go to your links, given that you're spamming our website. How freaking stupid can you GET?
The A15 is targeted at everything from phones to servers. Not only its cores are 25-40% faster than A9 cores at the same speed, they will still be very efficient. The A15 will be available from 1C@1GHz to [email protected], multiple processors can be linked together (AMBA® 4 Cache Coherent Interconnect), and it will have (LPAE) extensions for up to 1TB of RAM support (one of the main advantage of 64-bit architecture). It won't compete in raw performance with 2012 Core ix cpus, but it will still be king in terms of power consumption. Don't know for sure what will Apple do with it or when, but expect huge performance improvements on current designs and maybe good enough performance for new ones.
http://www.arm.com/products/processo...cortex-a15.php
not entirely true