The Apple iPad - Anand's Analysis

The Hardware

Leading up to today’s announcement I desperately tried to figure out what hardware Apple would use for the iPad. I’ve been on a bit of an SoC kick as of late, so you can understand my fascination.

Apple acquired PA Semi back in 2008. Everyone assumed that it’s because Apple wants to start making its own SoCs for the iPhone. Well, the first results of that acquisition are in the iPad.

Apple didn’t devote much time to the SoC in the iPad other than to tell the world that it’s Apple’s own silicon and it runs at 1GHz. The SoC is called the A4. I’ve asked Apple for more details on it, but I’m not holding my breath for a response.

Given the fact that it runs the iPhone OS and nearly all iPhone apps, I’m guessing the A4 is ARMv7 based. It’s possible that Apple engineered its own architecture for the A4, but more likely that it simply took an existing ARM design and modified it to suit its needs.

If Apple wanted to save cost it would’ve gone with a Cortex A8 based processor, or if it wanted more performance it would be something more A9 like. I’m not ruling out a dual-core implementation, but given the entry level cost point I’m assuming that it’s not anything quite as fantastic.

The 1GHz operating frequency implies a 45nm manufacturing process if it’s indeed an A8 or A9-like core. If we look at Apple’s public video, it appears to render a page at Spin.com in roughly 2.7 seconds. My iPhone 3GS does the same in about 7 - 9 seconds, but it also appears to be loading a lot more content on the current Spin.com site. Even if we assume that the 600MHz Cortex A8 in the iPhone 3GS can render the same page in 5 seconds, the speedup is too big to be from a clock speed increase alone.

Based on this data alone (and the responsiveness of the UI from the videos) I’m going to say that there’s a good chance that the A4 is much closer to the A9 in terms of performance. If it’s not an A9 itself, it may be Apple’s own out-of-order design.

Then there’s battery life. Apple is claiming 10 hours of web browsing battery life, which is reasonable given the 25WHr battery, but over a month of standby power. I suspect that the ridiculous standby power is due to the fact that the 3G radio is completely shut off when the device is asleep, but even then that’s very good power consumption. If anything, Apple’s own engineering here was probably spent on making sure that the SoC’s power consumption was as low as possible. By comparison, even the best SoCs in a smartphone today can usually only offer 300 hours of standby power (12.5 days).

Apple’s battery life claims have been unusually reliable as of late, so I would say that we should expect 10 hours of useful battery life out of one of these things.

I’ve spent a lot of time talking about the CPU, but what about the GPU in the A4? Given that Apple is a shareholder in Imagination Technologies (9.5%), I’d say that it’s a pretty safe bet to assume there’s some sort of a PowerVR SGX core in here. Which core? There’s definitely the physical space to include something ridiculous, but I’m guessing it’s something relatively controlled - perhaps an SGX 535 or SGX 540 at the most.

A4 in the next iPhone?

I've been racking my brain over the past several months trying to figure out what Apple will use in the next iPhone. I figured it could be as simple as a 45nm Cortex A8 shrink, or as ridiculously sweet as a pair of Cortex A9s. With the iPad being based on Apple's own A4 SoC design, I'm guessing we'll see it (or a derivative) making an appearance in the 4th generation iPhone.