Another reason you don’t want an Atom tablet; it might look something like this.

Recently some folks have voiced their opinion on the upcoming Apple Tablet, whatever it may be called. Gruber approaches the device from a practical but abstract “What will it be used for” perspective, while Siracusa looks at recent announcements to predict tablet features. Both approaches have their merit, but I think we can go a step further in looking at recent developments without getting into the wizzardy “analyst” prediction stage.

The iPhone and other smartphones

The iPhone 3GS is getting old in some ways, most notably its pixel density. While text on the iPhone still looks decent when compared to the competition, it’s no longer as superior as it once was. Mostly this is because newer smartphones have a higher pixel density. It would make sense for Apple to do the same. However, as so far all iPhones and iPod touches have had the same resolution, some applications will break when displayed on a different resolution. Of course, a Tablet running the iPhone OS will also have a different resolution than the current iPhones. It is therefore time for Apple to let their developers adapt to different resolutions, preferably before the next iPhone with a different resolutions is released, so that most existing applications will be ready for the change.

iPhone OS 4.0

So far, Apple’s iPhone releases have always coincided with new software releases. This makes sense on some level, as new features (with new API’s) on the phone can be combined with the updated OS, resulting in a single API change every year.

However, Apple has been critiqued for having a slow release schedule. Releasing a single phone every year is very sparse for a phone manufacturer and Apple isn’t currently planning on releasing a wide variety of different iPhones just to support the crazy phone market. They also don’t have the resources to release a quality update to the iPhone more than once a year.

One way they can resolve this problem is by decoupling the OS from the phone. The OS itself often brings new features (like the App Store, Spotlight of Voice Memos). Releasing an OS update once a year and a phone update once a year effectively means they have an updated phone to sell twice a year.

If Apple would do this, now would be the time to release a preview of iPhone OS 4.0. This new release would also be perfect for the support of different resolutions, perhaps combined with a new simulator that can simulate different resolutions. It could also create a solid foundation for the Tablet, which is important as Apple will want a wide variety of Apps on the Tablet at launch. The iPhone without custom applications was acceptable in 2006 (it probably isn’t anymore), but people will expect more from the tablet so application support right from the launch is important.

The Tablet release schedule

Current rumours point to a first preview at the end of January and a definite release somewhere in March. This means Apple will need to have something to demonstrate in January, but they still have time to ramp up production of components and tweak the software necessary. What could Apple demonstrate in January and have ready in March?

One thing is obviously an update to the iPhone OS. Apple could release an OS 4.0 Beta in January, then ship the Tablet in March or later with support for iPhone OS 4.0 apps. But I think that Apple can also have the hardware almost ready, though perhaps not a production capacity.

The ARM Cortex A9

In September 2009, ARM announced the Cortex A9 implementation. The Cortex A9 is the successor to the Cortex A8 currently used in the iPhone and most other smartphones. If Apple is to use the iPhone OS and expect the applications for the iPhone to work on the Tablet, they need to use an ARM processor. The only real alternative is the Intel Atom, which would require all developers to create universal binaries for their apps.

I don’t think this is very likely. Universal Binaries for OS X saw a relatively slow uptake of several months to a year, which Apple would like to avoid when announcing a new product. Also, the Intel Atom is frankly just not as attractive as the Cortex A9.

For one thing, the power / performance ratio is terrible. Furthermore the Atom takes up 3x as much space as the Cortex, allowing chip manufacturers to use a smaller footprint, or add more cores on the same die.

But that isn’t the biggest difference. The most important benefit of choosing ARM is that Apple can design its own chips. They have already bought PA Semi, which is rumored to be an ARM licensee. If PA Semi is capable of producing ARM chips, Apple could have a huge advantage. While the new generation of of Atom processors is also touted as being ‘system on chip’ (or SoC), this integration is weak when compared to the ARM processors. For example, the Pineview Atoms integrate video and memory controllers, but not important stuff like PCI, USB, SATA or audio buses. These buses can be made available by using the Intel NM10 chipset, but that also costs space, power, money and doesn’t even do HD video decoding, something even the chip used in the iPhone does.

Perhaps the biggest difference between the ARM licensing model and the Intel Atom model is that ARM licensees are allowed to add their own silicon to the chip design. Thus, Apple could, through PA Semi, create its own chip that does exactly what the Tablet needs: perhaps 2 or 3 Cortex A9 cores, a WLAN chip, audio, video, memory, USB controllers and HD decoders, but also stuff like multitouch controllers and hardware encryption support. This is a huge advantage above the Intel approach, where you basically need the processor and the chipset AND at least one other chip for your other peripherals. For a decent-sized tablet (something more portable than the Always Innovating Touch Book) chipset size is important. Plus, if Apple creates a custom chip, they can use derivatives of that for future iPhones — for example by using the low power Cortex chips or using less cores.

This of course raises the question of how soon the chips can become available. Traditionally, the process from reference implementation to an actually shipping product takes about 12 months. However, ARM has already demonstrated a Cortex A9 development board running at a lower speed than possible. The question is if PA Semi will be able to produce the required chip in time for the Tablet launch in high enough quantities. One important thing here is that if Apple is able to put chips on the market faster than the competition, they can have a huge advantage for the following years.

Conclusion

Summing up, Apple needs at least two things for a Tablet introduction. The January announcement would make sense in order to allow developers to adjust their apps for a March release. Apple might even be preparing for a surge in app submissions, as the average app approval time has been reduced to a few days. Some apps are review within 24 hours.

Next to that, the Tablet needs a good chipset. Having a bigger screen and more capabilities, common sense suggests it needs a more powerful chip than what’s currently in the iPhone. Apple should be very reluctant from abandoning ARM, with the current ARM roadmap and the large amount of App Store code already running on those processors. The Cortex A9 should be a good match for the tablet, rivalling the Atom chips at a lower power usage. The question is if Apple will be able to produce enough quality chips for a March release.