Implications of SGX543 in iPhone/Pod/Pad?

[3dcgi]

Cost is secondary?

They have the highest margins in the computer business. Probably the highest of the smart phone companies too.

They achieve lower costs through scale -- using A4 across 3 different product lines. Could they ramp up a high DPI display in high enough volume this year or next?

Apple has high margins because they make desirable products and often charge a premium for them. On top of that they have confidence in their products and this leads to their securing good component pricing early on. I don't know if they will go with a higher resolution display for the iPad 2, but it can use one and I'm sure they've considered it. ToTTenTranz says they wouldn't even consider it so I obviously disagree.

Do you honestly believe that? Have you ever seen an Apple keynote?
Half their speech is about price points.

Marketing and product development style are two different things. I don't believe Apple thinks "we need a $500 tablet" and designs a product to meet the price point. They say "we need a tablet people desire" and if the product will be too expensive for anyone to buy they don't make it.

The iPod was like this. Steve wanted the iPod before they made it, but Apple sat on the idea until the right storage device became available.

 

[Arun]

There's no better thread for this, so sorry if it's a bit off-topic but... let me make it official that I suspect the iPhone 5 will have a CSR SiRFstarIV GSD4t GPS chip. I'm far from certain from it, but I think it has become the most likely possibility now (with the integated GPS on the MDM6600 and the 65nm Broadcom chip as the second and third most likely options in my mind)

 

[tangey]

There's no better thread for this, so sorry if it's a bit off-topic but... let me make it official that I suspect the iPhone 5 will have a CSR SiRFstarIV GSD4t GPS chip.

In the same vein, seeing the block diagram of the TI OMAP5 and the inclusion on USB3.0, I was thinking that given that one of the major uses of iphone/itouch is music, that it would be a genuine user experience improvement for many if the next gen had USB3.0.

I know its started to appear on PC's this last 6-9 months, have any of the new Mac lines included USB3.0 ?

 

[Arun]

In the same vein, seeing the block diagram of the TI OMAP5 and the inclusion on USB3.0, I was thinking that given that one of the major uses of iphone/itouch is music, that it would be a genuine user experience improvement for many if the next gen had USB3.0.

USB 3.0 is a pretty big deal on handhelds IMO, and an even better use case than music is 1080p video (recorded your phone's camera to your PC - and from your PC to your phone). It's not strictly impossible that Apple supports it this year on the iPhone (after all their next-gen chip probably taped-out around the same time as the 40nm Marvell Armada 628 which does support USB 3.0) but I personally wouldn't expect it before next year.

 

[wco81]

I think they would want an installed based of USB 3 Macs before supporting it in their mobile devices.

It would be the logical next move, since they dropped Firewire altogether from their mobile devices and USB 3 is backwards-compatible.

But they some speculate that they're looking at LightPeak too.

They also have some agreement to support micro or mini USB for charging, per an EU requirement. So not sure what implications that might have for USB 3. They will keep the dock connector so any new IO would have to support that.

 

[Pressure]

I think they would want an installed based of USB 3 Macs before supporting it in their mobile devices.

It would be the logical next move, since they dropped Firewire altogether from their mobile devices and USB 3 is backwards-compatible.

But they some speculate that they're looking at LightPeak too.

They also have some agreement to support micro or mini USB for charging, per an EU requirement. So not sure what implications that might have for USB 3. They will keep the dock connector so any new IO would have to support that.

FireWire is still found in their offerings, just not the MacBook Air.

Chances of Apple adopting USB 3.0 is also slim if they can get LightPeak.

After all, the prototype computer running the LightPeak demo at its introduction were a Mac Pro sans the case.

The dealio with Europe and the "same power adapter to rule them all" is only voluntarily and not a legal requirement.

 

[wco81]

So we should be able to put to rest some rumors -- see which ones panned out -- this coming week.

Though in recent weeks, there are suggestions that what they release in the spring is an appetizer and the full meal comes in the fall. If there are really quad-core tablets with Tegra 3 and other vendors this fall, maybe Apple is preparing to rise to the game. Interesting that one of the things NVidia is touting is support for resolutions similar to the ones rumored for the iPad Retina Display.

Probably too early for ThunderBolt support in a tablet, though you can get USB 3.0 support through TB.

 

[Ailuros]

Apple resolution rumors have bounced back to the same iPad 1024*768 resolution in the meantime. Some even suggested a 7" display.

Since there are only 2 days left until the announcement I don't think it's worth dealing with that rumor mess.

On the SoC side rumors suggest 2*A9 CPU cores and a SGX543 MP2; given that it's most likely under 40nm it sounds feasible. If that's supposed to be the appetizer and in fall already there should be a "full meal" I'm not so sure 28nm would be ready by then, at least not in the mass quantities Apple would need.

 

[Entropy]

Apple resolution rumors have bounced back to the same iPad 1024*768 resolution in the meantime. Some even suggested a 7" display.

Since there are only 2 days left until the announcement I don't think it's worth dealing with that rumor mess.

On the SoC side rumors suggest 2*A9 CPU cores and a SGX543 MP2; given that it's most likely under 40nm it sounds feasible. If that's supposed to be the appetizer and in fall already there should be a "full meal" I'm not so sure 28nm would be ready by then, at least not in the mass quantities Apple would need.

40nm does seem reasonable, even though 32nm is an outsider option. Likewise, since Samsung leads with 32nm before offering a 28nm option in parallel, Apple may later go with either depending on timing, volumes and so on. I think it is a safe assumption that an unknown but significant portion of the 20000 wafers allocated to Apple by Samsung for 2011 will use finer lithography than 40nm.

It's not certain that March 2nd will bring us info on process technology used. That may have to wait a bit for third party investigation.

I have to wonder though - if Apple makes the leap from A8 and SGX535 to dual A9 and dual SGX543 on the same process node and with the same screen resolutions, what will they bring to the table when they have access to 28nm and/or choose to double the linear resolution for the iPad?

In particular I'm curious about where we're heading in terms of main memory technology for these devices, but maybe that should have it's own thread.

 

[Mike11]

I have to wonder though - if Apple makes the leap from A8 and SGX535 to dual A9 and dual SGX543 on the same process node and with the same screen resolutions, what will they bring to the table when they have access to 28nm and/or choose to double the linear resolution for the iPad?

Well, IMHO there's a small chance that a spring 2012 iPad 3 could have a retina display, A6 @28nm, 1 GB RAM, Cortex-A15 dual-core and POWERVR Series6. Quad-core Cortex-A9 and SGX544MP4 would be the "save" option I guess, at least for the iPad 3. IMHO at some point (2013?) Apple has to start developing two SoCs, one for the iPad and AppleTV (quad-cores?) and one for the iPhone and iPod touch (dual-cores?).

 

[Lazy8s]

I doubt 32/28nm will be ready for any Apple product this year. 543MP2 + A9 Dual will easily fit under the current process, and that'll compare well enough to Tegra 3.

Next year, or if they wanted to provide more performance for the TV and/or tablet products this year, they could slot in a couple more 543s. I see no reason why they'd suddenly switch to target the feature set of 544; performance is relatively similar.

I can't see Series 6 appearing anywhere before the second half of next year at the earliest.

 

[Ailuros]

I doubt 32/28nm will be ready for any Apple product this year. 543MP2 + A9 Dual will easily fit under the current process, and that'll compare well enough to Tegra 3.

In terms of SoC die area for sure.

Next year, or if they wanted to provide more performance for the TV and/or tablet products this year, they could slot in a couple more 543s. I see no reason why they'd suddenly switch to target the feature set of 544; performance is relatively similar.

I can't see Series 6 appearing anywhere before the second half of next year at the earliest.

Problem is that Apple keeps so tight lipped about their plans that a release has to be damn close to find out even just a few details. Their uber-next SoC on 28nm could either be some sort of "refresh" to the one announced in a couple of days or they could go for something way more aggressive both on the CPU as on the GPU side. Of course could the first scenario fit better their so far strategy but it's unfortunately not set in stone either.

4
I have to wonder though - if Apple makes the leap from A8 and SGX535 to dual A9 and dual SGX543 on the same process node and with the same screen resolutions, what will they bring to the table when they have access to 28nm and/or choose to double the linear resolution for the iPad?

I'm curious what changes they'll make over time in future iOS versions; just concentrating on the hw side is easy.

 

[Mike11]

I can't see Series 6 appearing anywhere before the second half of next year at the earliest.

Most likely you're right. But ST-Ericsson's Nova A9600 (dual-core Cortex-A15, Series 6) is scheduled to sample in 2011, and e.g. Nvidia's Tegra3 is announced to be available in tablets just 6 months after sampling. Everybody in the SoC game seems to be in a big hurry these days. And Apple is a special case anyway. But let's wait and see what Apple has to offer this generation first, maybe they disappoint us big time with just a higher-clocked A4

 

[rpg.314]

USB 3.0 is a pretty big deal on handhelds IMO, and an even better use case than music is 1080p video (recorded your phone's camera to your PC - and from your PC to your phone). It's not strictly impossible that Apple supports it this year on the iPhone (after all their next-gen chip probably taped-out around the same time as the 40nm Marvell Armada 628 which does support USB 3.0) but I personally wouldn't expect it before next year.

IIRC, John carmack recently said that flash in iPhone does ~30MBps. Assuming I remember correctly, and assuming those numbers are representative of that device class, it begs the question whether the storage subsystem of handhelds can saturate USB2.0, let alone USB3 or higher.

 

[metafor]

IIRC, John carmack recently said that flash in iPhone does ~30MBps. Assuming I remember correctly, and assuming those numbers are representative of that device class, it begs the question whether the storage subsystem of handhelds can saturate USB2.0, let alone USB3 or higher.

Well, flash performance can be increased (relatively) trivially by using more banks and a better controller. More importantly, a high-speed connector can open up other use-cases. For instance, streaming digital video may no longer require an HDMI connector, which I see as a potentially important feature for iDevices since they only have the dock connector.

 

[rpg.314]

Well, flash performance can be increased (relatively) trivially by using more banks and a better controller. More importantly, a high-speed connector can open up other use-cases. For instance, streaming digital video may no longer require an HDMI connector, which I see as a potentially important feature for iDevices since they only have the dock connector.

Using more banks may not be practical due to pcb size limitations.

PS: Can you have multiple banks within a single flash chip?

 

[metafor]

Using more banks may not be practical due to pcb size limitations.

PS: Can you have multiple banks within a single flash chip?

Even if you can, it'd require more I/O. I wonder if anyone's considered stacked flash as a possible solution. Are they even using both sides of the PCB to house the flash memory right now?

 

[Pressure]

Yeah, Apple is, at most, going to use a single chip in the iPhone and 2 in the iPad and iPod Touch.

 

[roninja]

Apple A5 announced inside the iPad2 - dual Cortex A9 + SGX543MP4 ? Or MP2??
9x graphics performance of A4 quoted.

 

[Lazy8s]

Definitely a possibility, though a 543MP2 could be described as providing a 9x boost depending on how much marketing was used.

Considering the conservative "up to 2x" quote for the general/CPU performance, a 543MP4 is realistic.