I/O & Overall Power

As we mentioned in the quick recap on the first page, the wording of Wolfson’s design loss at (presumably) Apple implies that they’ll likely still be in the 3G iPhone, so audio quality should be practically identical if that is indeed the case. The alternatives also wouldn’t provide substantially different quality, power consumption or cost, so it’s not even really worth discussing or speculating upon. Same for the power management & touch screen analogue as far as we can tell; it would be interesting if they got integrated into a single chip for cost reasons, but we do not believe that will likely be the case.

There’s another part of the input/output equation that’s potentially more meaningful here though: the screen. When playing video or browsing the web, the LCD screen represents a huge part of the system’s power consumption, and this can only get worse as newer chips become more power-efficient over time through Moore's Law and other means, unlike LCD backlights which aren’t really evolving all that much.

The alternative, of course, are OLED displays – while fully comparable raw data is hard to find, it seems that they are likely to about halve power consumption. Assuming a very highly efficient application processor and system architecture, this could result in nearly twice the video playback time: for VGA H.264 playback, some solutions are in the tens of milliwatts, while the screen may take hundreds. Web browsing and gaming would also take substantially less power.

On the other hand, music playback with the screen shutoff obviously wouldn’t be affected in any noticeable way. One possible way to think about music playback power consumption, assuming the lack of certain kinds of inefficiencies, is that it is equal to standby power consumption plus decoding plus audio analogue. With battery life for music playback in the ballpark of the original 2G iPhone, the MP3/AAC decoding hardware/software are likely a significant part of the equation; however, once again, through Moore's Law and algorithmic research it is becoming increasingly true that audio analogue (and system architecture) is a bigger limiter and blocks battery life from going substantially beyond 100 hours.

Overall, what we’re trying to point out here is that different workloads/usage scenarios have very different bottlenecks in terms of power consumption. It is perfectly possible to imagine a product with excellent video playback battery life but extremely subpar for music playback or web browsing. Furthermore, going from a 2G to 3G baseband should not significantly affect battery life for anything but phone calls and web browsing, and perhaps standby time. The web comparison is also rather unfair: would you rather have 5 hours of access where you need to wait 25 seconds per page, or 4 hours with refreshes that only take a few seconds?

In terms of costs, there is no reason to believe that connectivity wireless, the application processor or analogue is going to cost noticeably more than in the original iPhone. On the other hand, 3G is obviously going to add some extra costs compared to Infineon's 2G chipset, but these should not be overemphasized.