This is why Apple would simply use the multi-core version of the PowerVR currently inside the iPad, it is a product that the makers of the chip has ready for production in devices right now.
Do you have the power specs for the multi-core PowerVR parts vs the single core? multi-core isn't a free lunch, a dual core or a quad core vs a single core means exactly what it says, namely two or four times the number of cores which relates to twice or quadruple the number of transistors on the die. In fact it's usually more than double or quadruple because as the core count goes up the ancillary logic to handle bus arbitration and other coordination issues usually gets more complex. More transistors equals more switching current when active and more leakage current when inactive so typically the power consumption goes up. Often the hike in power consumption isn't linear with the number of cores because going multicore is usually done at the same time as rolling out a new process and/or geometry (e.g. 45nm to 32nm or smaller, HKMG, etc) but typically this improvement in fabrication does not fully offset the increased transistor count, especially in the first generation devices.
If Apple go to what they would likely declare as retina for the iPad by an exact quadrupling of the pixel count then that would imply a fourfold increase in the GPU performance, i.e. going to a quad core PowerVR for instance. I don't see that coming in at the same power profile as the current single core part, at least not for a year or two. If you have data that says otherwise then I'd be interested in seeing a link because I couldn't find power info on the multicore PowerVR stuff.
Something having more pixels does not mean it takes up more power. If there is more areas for a light diode to emit through then yes that would take up much more power.
This has also occured to me. I'm not an expert on displays but it did also seem to me that (excluding OLED) the big power drain is the backlighting and that should be proportional to the screen area and not the number of pixels. Like I say, I'm not an expert, but I tentatively agree with you on the above. All the points being made about the extra power needed to drive the substantially more powerful CPU and GPU needed to drive those extra pixels are still a big issue though I think.
As far as actually manufacturing a 9.7" screen with 4 times the pixel density, and with a total of over 5 times as many pixels as the iPhone 4 display, that isn't trivial. You do realise that in order to achieve the same yield as Apple are currently getting on the iPhone 4 display it's not a case of using a similar process and just scaling it up because if they did that then the current failure rate wouldn't be 5 times what it is with the iPhone, it would be the current iPhone failure rate raised to the 5th power. If the yield on the current iPhone is X (where X is between 0 and 1 with 0 being every single screen off the production line being a reject and 1 being no rejects at all coming off the line) then the same process scaled up to an iPad retina display would have a yield of X^5 (approximately, it's more like 5.12 to 2 d.p.).
I just don't think the technology is there yet, either in having that much CPU/GPU power within the same current draw, or being able to lay down that many pixels into a single panel with a low enough reject rate to make it economically viable. Ultimately we can argue this constantly until the day when Apple launch an iPad with a significantly higher ppi display, and in my view that day will definitely come, the only issue is when; I don't think that it's going to happen in 2011.
- Julian
