Next-Gen iPhone & iPhone Nano Speculation
- Thread starter Arun
- Start date
Possibly, both options are available and not contradictory to each other. The main thing is that it's clear that Apple is getting everything in line for console like playing on mobile and TV screens.
But it seems reasonable that they will their mobile devices lead in terms of graphics - both due to pricing constraints for the AppleTV, and due to volume. 300 million mobile iOS devices per year and still growing is a more powerful attractor for developers than a really powerful ATV. Once the concept is proven and the content is available, nothing prevents Apple from giving the ATV better stand alone gaming capabilities.
anexanhume
Veteran
It would be interesting to see them throw 16GB/32GB and a 6630 in one for $200 to $250 and see what developers do with it with the controller API.
We've had leaked front and backs, leaked trays of batterys, leaked plastic backs and multi-coloured options for the leaked name "5c", leaked champagne colour for 5s. Leaked internal main board, and various interconnects. Leaked dual flash led courtesy of the shape of the hole in the leaked casing.
In fact at this stage, just about every single aspect of the iphone 5s and 5c has been leaked.
Except of course for the one major differentiator and external change that has been assumed now for months, with just about every major news outlet taking for granted that there will be a finger print sensor built into the main button, and many citing this will be convex instead of concave.
So my question is, with every other part pouring out from multiple sources, how has apple managed to keep this single most important part from leaking anywhere, or does it really exist ?
In fact at this stage, just about every single aspect of the iphone 5s and 5c has been leaked.
Except of course for the one major differentiator and external change that has been assumed now for months, with just about every major news outlet taking for granted that there will be a finger print sensor built into the main button, and many citing this will be convex instead of concave.
So my question is, with every other part pouring out from multiple sources, how has apple managed to keep this single most important part from leaking anywhere, or does it really exist ?
My wild guess for the SoCs is a 375 MHz G643x variant in the A7 for the new iPhone and a 600 MHz G643x variant in the A7X for the new iPad. I can't figure what Apple might do with the CPU -- run the cores on separate voltage/frequency planes, add low power cores, add more high power Swifts, possibly some combination of the aforementioned options, -- but my guess for the clock speed of the high-end Swift rev. 2 cores is 1.5 GHz for the A7 in the iPhone and 1.8 GHz for the A7X in the iPad.
The early betas of iOS 7 was suppose run real well on iPhone 5.
Can't imagine how it would run on upgraded hardware.
Not that they have more performance than they need but it seems they might as well up the resolution and screen size if they're really putting that kind of GPU.
Has any device with Rogue been released yet?
Can't imagine how it would run on upgraded hardware.
Not that they have more performance than they need but it seems they might as well up the resolution and screen size if they're really putting that kind of GPU.
Has any device with Rogue been released yet?
Nebuchadnezzar
Legend
Whats the point in having two models if all the difference would be the clocks? Unreasonable in my opinion.
I'm just specifying the GPU and CPU cores in my guess; there's a whole lot that can vary between two SoCs using the same main cores obviously. There would be quite a few differences in other aspects of the SoC, even in the way those cores themselves might be configured as well as the supporting fabric into which they're implemented.
I just feel the G64xx should be within the die budget of a phone SoC, so Apple would favor it with the "wider is better" approach. If the G6630 were available in time, I'd say G643x for iPhone and G663x for iPad, but I think they'll make due with what they have.
I'm just taking a shot in the dark here, though. I've heard an equally reasonable proposal for a G62xx for iPhone and G64xx for iPad, so I don't feel any one option is obvious right now in contrast to how several of us were keyed in to the choice of 543MP2 for the A5/A5X a few months before any proof or leaks had surfaced.
I just feel the G64xx should be within the die budget of a phone SoC, so Apple would favor it with the "wider is better" approach. If the G6630 were available in time, I'd say G643x for iPhone and G663x for iPad, but I think they'll make due with what they have.
I'm just taking a shot in the dark here, though. I've heard an equally reasonable proposal for a G62xx for iPhone and G64xx for iPad, so I don't feel any one option is obvious right now in contrast to how several of us were keyed in to the choice of 543MP2 for the A5/A5X a few months before any proof or leaks had surfaced.
Brilliantdeve
Newcomer
what I found interesting though, some rumors website out there indicating that both Apple A7 & A7X might based ARMv8 .. 
ltcommander.data
Regular
http://stevecheney.com/on-the-future-of-ios-and-android/
NFC hasn't really taken off and with iBeacons in iOS 7, the thought is that Apple has found a better way to accomplish the same things and more. Apple won't likely be supporting NFC.
Well the A7X could still have the doubled memory controller to feed the higher clocked GPU.
I think Apple will go big die as well. The 32 nm A6/A6X have a smaller die area than the 45 nm A5/A5X and so Apple doesn't have to limit their transistor count by what going from 32 nm to 28 nm within the A6/A6X die area gives them, but can scale toward 45 nm A5/A5X die size. The iPhone 4S/iPad 2 claimed 2x CPU and 7-9x GPU over the previous gen and the iPhone 5/iPad 4 claimed 2x CPU and 2x GPU over the previous gen. I'm tending toward the iPhone 5S/iPad 5 having a 3x CPU and 3x GPU speedup over the previous gen. The implementation could be 1.6 Ghz/1.8 Ghz quad core Swift 2 with ~20% improved IPC and a 400 MHz G6430 and a 450 Mhz G6630 for the A7 and A7X respectively.
Swift seems to be focused on maximizing usage of the existing Cortex A9 execution units rather than adding more execution units, by adding a 3rd instruction decoder and giving each execution unit its own port. Swift 2 will probably be along the same vain. Besides the usual increases in reorder buffer sizes and improvements in branch prediction, IPC improvements if they are not already in Swift could come from the addition of L1 and L2 data predictors, dedicated integer divider, 128-bit NEON units and data paths, out-of-order execution for FP/NEON/load ops, and transitioning the loop buffer to after the decoders so the decoders can be powered down. If Apple moves to a quad core CPU hopefully they'll implement some form of independent power/voltage/frequency planes or gates able to completely shut down individual cores and frequency boost the remaining cores. They may also need to add a 2nd load/store unit like Cortex A12/A15.
nVidia claims not having memory disambiguation is an advantage, because it isn't power efficient. I wonder if that is really the case? As well, newer Cortex designs use dedicated issue queues for integer, FP/NEON, and memory ops while Intel sticks with a unified scheduler. Although we probably wouldn't know if they did, it'll be interesting to see if Apple adopted dedicated queues as well.
Apple may stick with 1GB of RAM, but hopefully they'll adopt faster flash memory so reloading apps is faster.
There certainly are enough changes in iOS 7 for Apple and devs to deal with without going 64-bit. Beyond simply building a 64-bit capable installed base for a future 64-bit iOS, ARMv8 standardizes some cryptographic acceleration which presumably would still be available in 32-bit mode and would be useful in combination with the fingerprint sensor. Of course previous SoC already had dedicated hardware encryption logic and Armv8 may not have been finalized in time for the A7/A7X.what I found interesting though, some rumors website out there indicating that both Apple A7 & A7X might based ARMv8 ..
Ehh, I certainly pray that if Apple has the option of going quad core, they'd instead spend the die area and thermal headroom on a beefier dual core and faster GPU.
Browsing is largely held back by single-threaded perf (stupid Javascript), and most creative/gaming apps are held back by GPU perf. The remaining folks who need more for niche apps should probably just get a notebook.
ltcommander.data
Regular
I'd be supportive on focusing on faster dual core CPUs and GPUs as well, but with other high-end smartphones standardizing on quad core CPUs there's going to be pressure on Apple to adopt more cores as well. Of course out of all the device manufacturers, Apple stands the best chance of convincing customers that more CPU cores isn't automatically better within the constraints of an overall device.
Apple's next-gen 'iPhone 5S' to use 31% faster 'A7' chip, feature motion tracking... http://appleinsider.com/articles/13...use-31-faster-a7-chip-feature-motion-tracking
anexanhume
Veteran
I simply don't think they've had the development time. Maybe next year.