Next-Gen iPhone & iPhone Nano Speculation

[wco81]

Rumors of iPhone in September, possibly along with iPad Mini in the same time frame.

iPod Touch is also due for a refresh and there are rumors it will get the larger screen as the iPhone.

Of course iPhone is going to get the priority as far as supplies of prime components. So perhaps older SOCs for the iPad Mini and iPod Touch?

Will be interesting to see how they slot the iPad Mini and iPod Touch pricing-wise.

 

[anexanhume]

Considering Apple has had silicon back on a Rogue equipped SoC for at least a little while and considering ST-Ericsson's miss on their projection of a late 2012 target for A9600 was a fab issue and not directly a SoC issue, I'm pretty sure Apple could be ready with a Rogue based GPU if they really wanted to push it for a 2012 iPhone. The A15 is the processor that'd give me some hesitation for an October 2012 projection, assuming Apple won't have their own custom ARM CPU solution.

I think Apple may wait to introduce Rogue with their new iPad, regardless, to benefit from better manufacturing process maturity.

Source on Apple having Rogue silicon? I don't see why they couldn't have A15 silicon too seeing as how Samsung is allegedly manufacturing their own 5250 right now (http://www.anandtech.com/show/5467/samsung-exynos-5250-begins-sampling-mass-production-in-q2-2012).

Rumors of iPhone in September, possibly along with iPad Mini in the same time frame.

iPod Touch is also due for a refresh and there are rumors it will get the larger screen as the iPhone.

Of course iPhone is going to get the priority as far as supplies of prime components. So perhaps older SOCs for the iPad Mini and iPod Touch?

Will be interesting to see how they slot the iPad Mini and iPod Touch pricing-wise.

32nm A5 for iPad mini and iPod Touch, I would guess.

 

[RudeCurve]

 

[ltcommander.data]

32nm A5 for iPad mini and iPod Touch, I would guess.

32nm A5 does seem to be a shoo-in. I do wonder if Apple will cripple them for the purposes of product differentiation. Something like the iPad Mini A5 at iPhones 4S 800MHz dual core rather than iPad 2 1GHz and iPod Touch as a single core A5 like the Apple TV. I hope not. Certainly keeping the iPad 2 and iPhone 4S CPU configurations for the iPad Mini and iPod Touch respectively would be best for software compatibility. Although iPod Touch apps will need some work regardless due to the expected resolution difference.

For the iPad Mini, although there aren't really a lot of rumours on this topic, I'm guessing they'll be also be available in cellular configurations. If I'm not mistaken, the cellular models have higher profit margins so would help amortize the design cost of the iPad Mini. If there are cellular models, what are the chances that the iPad Mini will have LTE? The iPad Mini could use the same 28nm baseband that the iPhone 5 is expected to use to conserve battery life despite not having the huge battery of the iPad 3.

With the iPhone 4S and iPad 3 doing 1080p video from the rear camera, it's very likely the new iPod Touch will be getting a 1080p rear camera as well, albeit fixed focus. With FaceTime over 3G coming with iOS 6, I'm thinking 720p FaceTime HD will be also be added in the iPhone 5 and iPod Touch. Omnivision seems to be offering 2MP still/1080p video (OV2722) and 720p cameras (OV9724 or OV9740) at similar ~3mm thicknesses to the existing cameras in the iPod Touch so it's definitely doable. The iPad Mini could get the same cameras as the iPod Touch.

I'm thinking Apple won't be interested in going all the way down to the $199 budget market and 8GB of flash memory is too small now anyways due to Retina iPad assets in apps, so I think $299 for a 16GB iPad Mini is a good starting point. A $299 16GB 7.85" iPad Mini should be competitive with the $249 16GB Nexus 7. $50 more for ~40% more screen area, albeit with reduced resolution, a rear camera, and the larger iTunes/App Store ecosystem.

I'm not sure where that leaves the $399 16GB iPad 2 though. A $299 16GB 7.85" iPad Mini with 32nm 1GHz dual core A5, 512MB RAM, 2MP/1080p rear camera, 720p front camera, a 1024x768 IPS screen, and LTE option, presuming similar battery life, would be superior to the $399 iPad 2 other than screen size. Apple will presumably be offering a 32GB iPad Mini for $399 as well. Will Apple try to make them more comparable, by for example retaining the 720p rear camera, VGA front camera for the iPad Mini, so both the iPad Mini and iPad 2 can sell alongside each other? Will the iPad Mini completely replace the iPad 2? Given Apple repeatedly pointing out the popularity of the $399 iPad 2 for education, I could see them doing as they do when releasing radical changes to Macs, which is keeping the $399 iPad 2 on as an education market exclusive since the larger screen size may be important for small children to accommodate their less co-ordinated gestures. Of course, if the iPad 2 is being kept around anyways for education, why not just let everyone have the option like what happened to the eMac. It's only the iPad 2 that has this ambiguity with the iPad Mini. Next year if/when the iPad 3 falls to $399, the Retina display can be used as the clear differentiating factor between the full iPad and the iPad Mini.

EDIT:
http://www.macrumors.com/2012/07/30...irmed-for-mid-september-iphone-launch-likely/

September 12 Apple event confirmed.

 

[wco81]

Although iPod Touch apps will need some work regardless due to the expected resolution difference.

iPod Touch already has a Retina display, though it may not be IPS quality.

A cell modem makes more sense with the smaller iPad, though if they keep the $130 price delta over the Wifi-only SKUs, that's a bigger bite relatively speaking.

 

[ltcommander.data]

iPod Touch already has a Retina display, though it may not be IPS quality.

It has the current 960x640 resolution, but it should be moving to 1136x640 @ 4" like the iPhone 5. That'll make a 1136x640 A5 5th gen iPod Touch unique versus a 1136x640 A6 iPhone 5 and the 960x640 A5 iPhone 4S. It does make you wonder whether Apple will bump up the A5 clock speed in the 5th Touch from 800Mhz to 1GHz to account for the resolution increase and maintain roughly the same processing power per pixel as the iPhone 4S. The need to save 1GHz parts for the iPad Mini and the increased power consumption may prevent that though.

 

[wco81]

Oh right, the rumored higher resolution and big screen.

Thing is, you would also expect them to continue making iPhone 4 and iPhone 4S at reduced prices even after they announce the new iPhone. So they have to keep producing a number of A4 and A5 chips for those products, in addition to these rumored new products.

Plus they're trying to bring up TMSC as a secondary supplier of SOCs, given they're they're going into a big lawsuit against Samsung?

 

[anexanhume]

iPod Touch already has a Retina display, though it may not be IPS quality.

A cell modem makes more sense with the smaller iPad, though if they keep the $130 price delta over the Wifi-only SKUs, that's a bigger bite relatively speaking.

Reviews have noted it doesn't have the same viewing angles, suggesting it's TN versus IPS.

Oh right, the rumored higher resolution and big screen.

Thing is, you would also expect them to continue making iPhone 4 and iPhone 4S at reduced prices even after they announce the new iPhone. So they have to keep producing a number of A4 and A5 chips for those products, in addition to these rumored new products.

Plus they're trying to bring up TMSC as a secondary supplier of SOCs, given they're they're going into a big lawsuit against Samsung?

Bringing TSMC as another source of SoCs wouldn't be due to their legal battles. Samsung microelectronics is a completely separate division from their mobile one, and Apple is a huge cash cow they wouldn't turn down. Furthermore, it seems unlikely they'd have multiple sources for any given SoC, given the added design time and cost associated with another foundry's process and standard library.

Also, it appears that Samsung's fab that supplies Apple is in Texas, not Taiwan. I'm not sure if that's a motivator for Apple, but Samsung is a lot less of a roller coaster because they don't have a lot of customers like TSMC with unpredictable demand cycles. 28nm has been a rough road for them, just as 40nm was. Qualcomm is already racing to UMC to get more 28nm fab capacity for their Krait chips, but also likely their new 28nm LTE radios which are probably going to end up in the next iPhone and iPad.

 

[ltcommander.data]

Bringing TSMC as another source of SoCs wouldn't be due to their legal battles. Samsung microelectronics is a completely separate division from their mobile one, and Apple is a huge cash cow they wouldn't turn down. Furthermore, it seems unlikely they'd have multiple sources for any given SoC, given the added design time and cost associated with another foundry's process and standard library.

Also, it appears that Samsung's fab that supplies Apple is in Texas, not Taiwan. I'm not sure if that's a motivator for Apple, but Samsung is a lot less of a roller coaster because they don't have a lot of customers like TSMC with unpredictable demand cycles. 28nm has been a rough road for them, just as 40nm was. Qualcomm is already racing to UMC to get more 28nm fab capacity for their Krait chips, but also likely their new 28nm LTE radios which are probably going to end up in the next iPhone and iPad.

On the subject of Samsung and 28nm processes, how close is Samsung's 28nm process to mass production? Sometime in 2013? They describe it as an easy migration from 32nm and the 28nm LPH process offers a nice reduction in active power (20%) or a boost in performance (20%). The iPhone 5 getting an incremental SoC on existing technology (high clock speed Cortex A9 + SGX543MP) or with only one subsystem upgraded (Cortex A15 or Rogue) at 32nm, followed by the iPad 4 introducing Cortex A15 + Rogue at 32nm in H1 2013, and then the iPhone 6 getting a 28nm shrunk version in H2 2013 seems like a sensible progression.

 

[anexanhume]

On the subject of Samsung and 28nm processes, how close is Samsung's 28nm process to mass production? Sometime in 2013? They describe it as an easy migration from 32nm and the 28nm LPH process offers a nice reduction in active power (20%) or a boost in performance (20%). The iPhone 5 getting an incremental SoC on existing technology (high clock speed Cortex A9 + SGX543MP) or with only one subsystem upgraded (Cortex A15 or Rogue) at 32nm, followed by the iPad 4 introducing Cortex A15 + Rogue at 32nm in H1 2013, and then the iPhone 6 getting a 28nm shrunk version in H2 2013 seems like a sensible progression.

Supposedly Samsung 28nm capability is already ramping http://www.youtube.com/watch?v=iMNp1yK7bmE , but I had head nothing about 20% gains. I'd expect around 12% really. However, Apple would definitely be interested at 20% for sure. I can see the next iPhone going a lot of places, but I certainly expect the next iPad to have A15, Rogue and 32nm/28nm process. There's a lot of stepping stone possibilities in between for sure.

 

[Jubei]

Didnt Qualcomm secure a contract with Samsung for 28 nm recently? If true then they are ready for mass production. I just dont see a company like Apple risking their yield rates when a perfectly stabile process is available at 32nm. They released those die shrunk iPads for a reason

Would have to agree that iPhone 6 would be the earliest we will see Apple at 28nm. Probably do another test drive with iPad 3 in lesser volumes

 

[ltcommander.data]

Supposedly Samsung 28nm capability is already ramping http://www.youtube.com/watch?v=iMNp1yK7bmE , but I had head nothing about 20% gains. I'd expect around 12% really. However, Apple would definitely be interested at 20% for sure. I can see the next iPhone going a lot of places, but I certainly expect the next iPad to have A15, Rogue and 32nm/28nm process. There's a lot of stepping stone possibilities in between for sure.

http://www.samsung.com/us/business/oem-solutions/memory-logic/foundry/foundry-32nm.html
http://www.samsung.com/us/business/oem-solutions/memory-logic/foundry/foundry-28nm.html

Samsung's claiming the 32nm process provides 41% less active power or 27% more performance at the same leakge versus their 45nm process. They report 60% less active power or 55% more performance for their 28nm LPH process versus their 45nm process. So that's about a 20% power reduction or 20% performance boost when comparing 28nm to 32nm.

http://www.anandtech.com/show/6126/glbenchmark-25-performance-on-ios-and-android-devices/2

Anand's GLBenchmark 2.5 comparison shows Adreno 225, Tegra 3, and high-clock speed Mali-400MP4 have largely caught up or overtaken the SGX543MP2 in the A5, notably in the forward looking Egypt HD. If Apple does intend to focus on holding the clear GPU crown they'll need to be pretty aggressive. Even a 50% clock speed bump wouldn't provide much performance separation from the competition.

 

[ams]

Yes, Geforce ULP and Adreno have relatively strong geometry performance. That said, [SGX543MP2] GPU clock frequency will be nearly double on iphone 5 vs. iphone 4s, so no doubt the GPU performance will be very good in comparison. The HTC One X international version (with quad core A9 CPU / Tegra 3) and Samsung Galaxy SIII international version (with quad core A9 CPU) will have stronger CPU performance than iphone 5 (with dual core A9 CPU).

 

[Pressure]

Are we all ignoring that Apple still has the fastest SoC from the GPU standpoint in the A5X? But it's good the competition has finally caught up (with the A5).

I wouldn't be surprised if they simply shrinked the A5X to 32 nm or 28 nm. Or traded the Cortex-A9 in for Cortex-A15 and called it a day.

People seem to think Rogue is still a bit early for a 2012 release.

 

[Jubei]

Apple doesnt really need to beat the Android alliance in GPU power. They are only competing with themselves. So doubling the clock speed would allow them to claim 2x as powerful as iPhone 4S in their marketing wich is all they really need anyway

With that said i still dont believe the next chip will be a simple shrink with higher frequencies. The codenames suggest a new gen rather than a small upgrades

 

[Ailuros]

Apple doesnt really need to beat the Android alliance in GPU power. They are only competing with themselves. So doubling the clock speed would allow them to claim 2x as powerful as iPhone 4S in their marketing wich is all they really need anyway

With that said i still dont believe the next chip will be a simple shrink with higher frequencies. The codenames suggest a new gen rather than a small upgrades

Doubling frequency might be a bad idea in terms of power consumption vs. for example more GPU cores. I'd suggest that a hypothetical MP4@200MHz would give them roughly 70% GPU performance of a iPad3.

People seem to think Rogue is still a bit early for a 2012 release.

To be honest I wouldn't be worried with someone like Apple about die area, which would be my primary consideration for early Rogue integration for everyone else. The reason my gut feeling tells me that Apple won't show up as early as 2012 with a SoC containing a Rogue GPU, is that I doubt it would be ready for integration that early, even more so considering Apple's volumes which means that they're probably stock piling for quite some time before each hard launch.

 

[anexanhume]

Apple doesnt really need to beat the Android alliance in GPU power. They are only competing with themselves. So doubling the clock speed would allow them to claim 2x as powerful as iPhone 4S in their marketing wich is all they really need anyway

With that said i still dont believe the next chip will be a simple shrink with higher frequencies. The codenames suggest a new gen rather than a small upgrades

We don't know what apple considers constituting a new rev though. They could just go quad A9.

 

[Jubei]

We don't know what apple considers constituting a new rev though. They could just go quad A9.

Anything is possible but i think quad A9 is unlikely. iPhone is not really that CPU dependant and there are no apps (unlike the iPad) that could utilize 4 cores so imo it would be a waste of die space. At the same time given how conservative Apple is with power draw i cant see them using Cortex A15 on its own as i believe it draws more at idle than Cortex A9 and i dont believe big.little technology will be available this year. Maybe they would clock it at 600 MHz or something.

 

[anexanhume]

Anything is possible but i think quad A9 is unlikely. iPhone is not really that CPU dependant and there are no apps (unlike the iPad) that could utilize 4 cores so imo it would be a waste of die space. At the same time given how conservative Apple is with power draw i cant see them using Cortex A15 on its own as i believe it draws more at idle than Cortex A9 and i dont believe big.little technology will be available this year. Maybe they would clock it at 600 MHz or something.

big.LITTLE is essentially just that A5 core being available. They're free to implement their own flavor like Nvidia did with the A8 core on Tegra 3.

 

[Jubei]

big.LITTLE is essentially just that A5 core being available. They're free to implement their own flavor like Nvidia did with the A8 core on Tegra 3.

Its a bit more than just that but i agree Apple does have their own ARM license and could essentially do whatever they felt they needed

the Tegra 3 shadow core is still an A9 just built on a LP process