Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple and TSMC Reportedly Completing Designs for 20-nm A7 Chip With Early 2014 Availability
- Thread starter MacRumors
- Start date
- Sort by reaction score
It's just hilarious to read grown (I assume) men critiquing a supplier's logo. It's exactly the kind of thing my little sister does with her friends about almost anything. She's at that age of "know what's trending" or else!
Of course, maybe the comments are from her friends...aaackkkk! Sorry, Becky.
So smaller means faster and more battery life?
Imagine if just for fun
put 2 Die shrunk A6s in the iPhone 5S.It would crush.
So do you think I should buy Apple stock when it drops below 100? An significantly improved processor is a major selling point for a new iPhone, especially the 'S' model. Take that away from the iPhone 5S and iPad 5...well, at least the iPad 5 will look/feel different, no such luck with the iPhone 5S. Apple knows they have to deliver the A7 this year, even if it's 28nm. If they don't, the 5S will get ripped apart by professional reviewers, even the Apple fanboy ones. Anyone who says updating this year makes no sense is an iPhone 5 owner who doesn't give a s*** about iPhone 4S owners months away from paying for essentially the same phone they just bought (if it still has the A6).
Apple and TSMC together at last? Maybe it's time to dump the TSM stock. Since I suggested it 2 years ago, it's had a pretty decent return, no?
Why dump the stock? This news sounds like more reason to buy. Even w/ a decent return, should we expect improved revenues with Apple supposedly moving more production to them?
Interesting. The smaller process could enable all the obvious benefits: lower power consumption, more clock-rate headroom, lower cost because more chips can be cut out of a single silicon wafer, and all that good stuff.
But maybe the biggest benefit is quad CPU cores on the A7. Just a thought.
----------
How about a 28-nm dual-core A7 in the iPhone 5S this year?
And a 20-nm quad-core A7 in the iPhone 6 next year?
Not good enough? Well have you tried switching to decaf?
But maybe the biggest benefit is quad CPU cores on the A7. Just a thought.
----------
How about a 28-nm dual-core A7 in the iPhone 5S this year?
And a 20-nm quad-core A7 in the iPhone 6 next year?
Not good enough? Well have you tried switching to decaf?
to be clear, it's usually the customer that tapes out to the chip vendor. it doesn't really make sense to say TSMC is taping out.
way back in the day the designs were written to magnetic tapes and sent to the fab, hence the terminology. we still use it today to mean the handoff between customer and foundry.
way back in the day the designs were written to magnetic tapes and sent to the fab, hence the terminology. we still use it today to mean the handoff between customer and foundry.
I need decaf and you need glasses. Did you read my post? I mentioned the possibility of a 28nm A7 for the 5S. I understand using the A6 for the iPad mini 2 and next version of Apple TV, but for the next iPhone and iPad, I would be surprised.
Id just assume keep a dual core A7 at 20nm and just have all that extra runtime on the battery.
While everybody seems to think 5S will come sooner rather than later this year, it may not. 5S may just as well arrive this fall, roughly a year after 5.
Shrinking A6 from 32nm to 28nm would not be done for the sake of the shrink, but to move from 32nm operating Samsung foundry to 28nm operating TSMC foundry. As others note, while the A6 is indeed an impressive chip, the 5S carries with it the expectation of A7 with a real performance gain. A6X of course makes zero sense in a phone, the X lineage is tablet-specific.
20nm TSMC process is not production ready and won't be well into early 2014 - it's a significant node jump from 28nm and TSMC does not posses a striking track record launching new processes lately.
New process is best tested with a known good chip design as Apple demonstrated with 32nm A5. Hence this 20nm A7 will most definitely be a die-shrink of a 28/32nm A7 that will land in 5S this year.
The question is, will Apple move to TSMC already with A7 or stay with Samsung? 28nm TSMC and 32nm Samsung processes are roughly equal and both well established. Apple just happens to have hands-on experience with Samsung, so this path would be easier. Then again Apple may want to test drive TSMC's fab ecosystem ahead of toying with the bleeding edge 20nm process next year. Other reason for the switch is to hide A7 design from Samsung's eyes. But why are we not hearing anything about a 28nm Apple design with TSMC?
Apple obviously still has some plans with Samsung's 32nm process as they demonstrated with the latest single-core significant rework of ATV's A5. Claimed to have interesting analog blocks at 32nm, not only the logic.
Talks about cheaper iPhone and the possible use of Qualcomm SoC there due to integrated cellular baseband - well, how about licensing the baseband IP from Qualcomm to integrate into Ax design? Baseband is pure digital, so it should be possible to bring to Samsung just as well as TSMC, although the latter would be an easier job as Qualcomm already have the chip layout on their 28nm process.
Possibilities endless...
Shrinking A6 from 32nm to 28nm would not be done for the sake of the shrink, but to move from 32nm operating Samsung foundry to 28nm operating TSMC foundry. As others note, while the A6 is indeed an impressive chip, the 5S carries with it the expectation of A7 with a real performance gain. A6X of course makes zero sense in a phone, the X lineage is tablet-specific.
20nm TSMC process is not production ready and won't be well into early 2014 - it's a significant node jump from 28nm and TSMC does not posses a striking track record launching new processes lately.
New process is best tested with a known good chip design as Apple demonstrated with 32nm A5. Hence this 20nm A7 will most definitely be a die-shrink of a 28/32nm A7 that will land in 5S this year.
The question is, will Apple move to TSMC already with A7 or stay with Samsung? 28nm TSMC and 32nm Samsung processes are roughly equal and both well established. Apple just happens to have hands-on experience with Samsung, so this path would be easier. Then again Apple may want to test drive TSMC's fab ecosystem ahead of toying with the bleeding edge 20nm process next year. Other reason for the switch is to hide A7 design from Samsung's eyes. But why are we not hearing anything about a 28nm Apple design with TSMC?
Apple obviously still has some plans with Samsung's 32nm process as they demonstrated with the latest single-core significant rework of ATV's A5. Claimed to have interesting analog blocks at 32nm, not only the logic.
Talks about cheaper iPhone and the possible use of Qualcomm SoC there due to integrated cellular baseband - well, how about licensing the baseband IP from Qualcomm to integrate into Ax design? Baseband is pure digital, so it should be possible to bring to Samsung just as well as TSMC, although the latter would be an easier job as Qualcomm already have the chip layout on their 28nm process.
Possibilities endless...
Your forgetting that the first quarter of 2014 is October of 2013. That is when the new iPhone whatever it is called will be introduced. So if true the A7 would be in the new iPhone.
But if you believe anything the Digitimes says I have beach front property i will sell you in death valley.
----------
If you want the gory details of the A6 here is your read:
http://www.anandtech.com/show/6330/the-iphone-5-review/22
If you want the gory details of the A6X here is your read:
http://www.anandtech.com/show/6472/ipad-4-late-2012-review/2
As for the iPhone 5S or whatever it will be named I would think it will get the A6X. In the resolution the iPhone runs it will kick ass for speed and graphics use because it wont have to drive the huge display of the iPad. Because the A6 in the iPhone 5 Apple adds a third PowerVR SGX 543 core and runs the three at a higher frequency than in the A5.
In the A6X Apple adds a Fourth PowerVR SGX 543 core and runs the Four at a higher frequency.
On an interesting note the New Galaxy S4 runs only 3 PowerVR SGX 543 cores. Same as the iPhone 5, which is the first time samsung has used PowerVR
Last edited:
Using your logic, why do you want to needlessly kill battery life to drive current through GPU transistors that are under utilized for the 4" LCD?
They won't if they go to 20nm process.
Also all A series chips are scalable for whatever processes you are running. In the articles quoted above Adnand La Shimpi explains that the A6 and A6X process things so fast that they use less power because they complete there processing tasks faster than other chips on the market and they clamp power usage to lower levels faster than any other mobile processors on the market when not in use for intensive processing, including stock off the self arm cores like the A9 or the A15's. By doing this they conserve power in a very fast and brutally quick way. And idle at lower power levels than others. This is because Apple custom builds there A processors instead of using the off the shelf speck from arm. They only use the ARMv7 guidelines not a stock ARM processor.
Last edited:
I think we still have a possibility of A7 on 28nm in late 2013... An early 2014 availability for A7 on 20nm will mean a product somewhere in late Spring/early summer given the manufacturing and ramp up cycle. it will be wait too long to let A6 fencing all the bad boy CPU from Samsung/Qcom etc for that long.
Really? Fast enough for a desktop? The A6 is still beat by 2004-2005 Desktop CPUs, how the hell would it replace an i7?
It wouldn't. a 2-4 chip but lower per chip cost version could be a Mac-Mini. A cripple-mac. It's how they roll.
This new chip may not be stamped out in Taiwan . It could come to Oregon or New York or Washington State.
http://www.oregonlive.com/silicon-f...oject_azalea_a_thre.html#incart_more_business
http://www.oregonlive.com/silicon-f...oject_azalea_a_thre.html#incart_more_business
Yes, a 2-4 chip set would still be far slower than an i5 or i7, but would increase profit margins! And screw compatibilty with 80% of software written.
Good point. Or maybe just an extra hour of battery life with a thinner battery.
Not that the '5' needs to get much thinner really.
----------
I stand corrected.
Any rumours on Apple licensing capacity off Intel? Global Foundries (formally the fab division of AMD) might also be a partner in the future given the rumours not too long ago regarding Apple employees around the AMD campus.