Next-Gen iPhone & iPhone Nano Speculation

[anexanhume]

For those who still like to believe in an A6 in the next iPhone:

Beware: it's a sketchy picture of a picture on a monitor...

http://9to5mac.com/2012/08/30/more-iphone-parts-a6-processor-more-new-9-pin-cables/

Edit:
There's some speculation that even if Apple's next SoC is called A6 it might still just be a higher clocked A5R2 (because it's just branding). But I don't think Apple would do that.

Luminance removed (not my photos)

ZOOM:

Notice text on other chips, QR code with high contrast not visible in these images. However, source said it had to be enhanced with photoshop to be visible, so it may just be that.

Oops, didn't mean to DP.

 

[Lazy8s]

A lot of tuning, especially on the process side, happens between getting early silicon back and going into volume production.

 

[anexanhume]

A lot of tuning, especially on the process side, happens between getting early silicon back and going into volume production.

I'm aware of that. I'm saying, what is the point of asking if there is enough volume? If the design is ready, it's ready. It's a design lead time issue. 32nm is fairly mature at this moment,

 

[Lazy8s]

... I must still be missing something... Even ignoring the issue of whether early silicon will equate to final silicon, the stage of ramping up to volume production is not simply a function of time.

 

[anexanhume]

... I must still be missing something... Even ignoring the issue of whether early silicon will equate to final silicon, the stage of ramping up to volume production is not simply a function of time.

Then what is the function?

 

[Lazy8s]

On the process side alone, it's the proficiency with which the engineers can determine an optimal recipe for implant/diffusion with the combination of UV bakes, radiation, gases, etc., among other fabrication stages, and whether any of that has to feed back into the design of the chip.

 

[anexanhume]

On the process side alone, it's the proficiency with which the engineers can determine an optimal recipe for implant/diffusion with the combination of UV bakes, radiation, gases, etc., among other fabrication stages, and whether any of that has to feed back into the design of the chip.

I'm specifically looking for things that would allow limited production but not volume production that isn't related to an immature process.

Also, what could be fed back into design that doesn't cause new million dollar masks to be made? All chips use the same process library, so what individual characteristics are driving those tweaks?

 

[silent_guy]

On the process side alone, it's the proficiency with which the engineers can determine an optimal recipe for implant/diffusion with the combination of UV bakes, radiation, gases, etc., among other fabrication stages, and whether any of that has to feed back into the design of the chip.

I have no idea what you're talking about.

Yes, the design house gives feedback to the fab. Like 'the yield is too low, fix it', but there is very little reverse feedback, other than to point out flaws in ATE coverage when there is abnormal yield loss. Once a base layer is frozen, what exactly do you expect the design house to do? Respin? That almost never happens... And if it happens it's to fix blunder logic bugs, not to work around some process issue. And it's not as if you can do a lot of process related fixes with a metal only fix either. The time for maximum communication from fab to designer is long before tape-out, not after.

The time between initial silicon and production is spent doing relentless testing of any parameter you can think of: let the chip to thousands of time to sleep and see if it wakes up correctly every single time, test OSD structures to make sure you don't zap the silicon on a production floor, long term reliability tests at high temperature, drift of key analog parameters over time, impact of noise on operation, high coverage ATE testing, ... If you're producing 100M chips per year, you have to shoot for extremely high standards.

 

[Lazy8s]

I stated "whether any of that (process tuning) has to feed back into the design of the chip", simply acknowledging the worst case scenario when considering the progression from early silicon to volume production (for example, a design that's too ambitious in its combination of embedded RAM and logic, requiring a great many cycles through the implant phase.) Tuning a manufacturing process certainly doesn't necessitate, nor even make likely, a respin, yet it is a time-consuming challenge that needs to be overcome when transitioning to volume production.

 

[wco81]

So is anyone shipping products with A15 and/or Rogue cores this year?

 

[Ailuros]

So is anyone shipping products with A15 and/or Rogue cores this year?

I don't think so, yet love being wrong.

 

[iwod]

So is anyone shipping products with A15 and/or Rogue cores this year?

Well it was supposed to be Sony Mobile coming in Xmas time frame. But they have pushed it back to a more realistic time frame in 1st half 2013. Which means in Real Life time we should be able to see Phone with big.LITTLE Cortex A15 and Rogue SoC same time next year.

Even the A6 is fake, if the pic is true it seems we will be getting a similar big size SoC as iPhone 4S.

So there is at least some hope of a faster SoC.

 

[Mike11]

Even the A6 is fake, if the pic is true it seems we will be getting a similar big size SoC as iPhone 4S.

Good point, but it's hard to tell how big the SoC is since everything around it changed. AFAIK the only constant might be the length of the SIM slot. The new SIM card tray is supposed to be as long as the old one, just not as wide.

The A5R2 is bigger than the A4 but of course smaller than the A5:

Here are the A4 and A5. I didn't try to measure it, but the new SoC seems to be at least bigger than the A4. Which tells us nothing

 

[anexanhume]

So is anyone shipping products with A15 and/or Rogue cores this year?

No one is shipping Rogue this year but both Samsung and TI have availability of A15 later this year according to road maps.

 

[Helmore]

No one is shipping Rogue this year but both Samsung and TI have availability of A15 later this year according to road maps.

There have already been some demos on show floors of TI's OMAP 5 SoCs and Samsung's Exynos 5250. The OMAP 5 demo wasn't at production ready frequencies though and there was really little info about the frequencies that the Exynos was running at during their demo. I think we can expect both of them to be ready by the end of the year, but I'm not sure whether we will actually see products that use them before the end of the year. And another thing, neither of them use Rogue, but they're both Cortex-A15 dual cores.

 

[anexanhume]

There have already been some demos on show floors of TI's OMAP 5 SoCs and Samsung's Exynos 5250. The OMAP 5 demo wasn't at production ready frequencies though and there was really little info about the frequencies that the Exynos was running at during their demo. I think we can expect both of them to be ready by the end of the year, but I'm not sure whether we will actually see products that use them before the end of the year. And another thing, neither of them use Rogue, but they're both Cortex-A15 dual cores.

Yes, and as far as I know, ST is the only one to have even announced a Rogue product.

 

[wco81]

Even RAM would help. This iPad 3 is constantly reloading tabs, tho otherwise snappy.

Speculation on the iPad Mini is A5, probably running on 32 nm, just like the new iPad 2?

At least in tablets, if lower price points drive higher volumes, then the adoption of newest SOCs for tablets may slow down.

 

[anexanhume]

Even RAM would help. This iPad 3 is constantly reloading tabs, tho otherwise snappy.

Speculation on the iPad Mini is A5, probably running on 32 nm, just like the new iPad 2?

At least in tablets, if lower price points drive higher volumes, then the adoption of newest SOCs for tablets may slow down.

Even with the iPad mini potential customers I think the market will exist for the "big kids" and the 500+ tablets.

 

[ltcommander.data]

Even with the iPad mini potential customers I think the market will exist for the "big kids" and the 500+ tablets.

And even if for tablets Apple starts producing a dedicated low-cost tablet rather than just reusing last year's models, they are likely still using as many parts from last year's model as possible. That keeps up pressure to invest and push technology for each year's top-tier model, because they have to aim high enough now so that the tech is still viable to reintroduce in a new lower-end product next year.

I do wonder, after all the costs and sales are added up, whether it'll work out that this year's $500+ tablets subsidizes next year's sub-$500 tablets or whether it'll be next year's sub-$500 tablets subsidizing this year's $500+ tablets. No doubt Apple aims for all their products to be profitable, but it'll be interesting to see which is the main profit center.

 

[iwod]

There have already been some demos on show floors of TI's OMAP 5 SoCs and Samsung's Exynos 5250. The OMAP 5 demo wasn't at production ready frequencies though and there was really little info about the frequencies that the Exynos was running at during their demo. I think we can expect both of them to be ready by the end of the year, but I'm not sure whether we will actually see products that use them before the end of the year. And another thing, neither of them use Rogue, but they're both Cortex-A15 dual cores.

What are its production freq anyway? 1.5Ghz? We dont need that. Dual Core A15 1Ghz would do for me