Originally Posted by RBR
A product generation is different than a generation of chip production and always will be.
Sure. But those products aren't going to be that different because it's all trim without the generational improvement in performance.
There will be different components that are incorporated into the product as they become available.
There is also the matter of new products, such as the iPad, using "old stuff" to get something out the door when product introduction timing is more important than waiting for the supposed current generation of something to become available in quantity.
The A4 in the iPad was the most modern SoC when it shipped in April 2010. 45 nm Cortex-A8 with a SGX535 GPU. And the IPS screen is the best kind of LCD screen available.
The iPad 2 (or whatever it will be called) should reflect what would have been expected of the first generation product, but was not done because of time constraints.
If the iPad 2 is a dual-core A9 with an SGX543MP, it could not have been shipped in 2010 in any kind of performance, cost benefit trade. It was not achievable. However, on mature 45 nm fabs with a good voltage bin or a half node improvement at 40nm, it's achievable.
Why Apple chose not to ship the iPad 1 with 512 MB RAM, I don't know. It could be as simple as they wanted to make their margin at $499, even though it would have been a $5 to $10 difference in component costs. Maybe it was planned obsolescence.
The shift of process node to a smaller one also has an impact on pricing of the product or margins. At some point the chip manufacturer is "selling silicon" and smaller is cheaper.
Well, yes, but it also makes higher performance chips possible: double the transistors, lower voltages, higher frequencies. It's a Highlander-ish competition really.