OK, I'll meet you halfway - let's try manufacturing economics - and then I'll drop it, because I specialize in wafer production for a living - it's what my company has done for more than the past decade (as an outgrowth of weapons research, for Namrak
).
~~~~~~~~
Case 1 - I have a wafer full of the same SoC, but I can simply change embedded firmware to decrease the capability of one (4460) to get the capability of the other (4430).
So - in that case, it costs me more to produce the 4430 but I'll charge you less.
Not reasonable.
Case 2 - The 4430 and 4460 are actually wired differently in their mask/layouts - yet, I have a product split on the wafer and whenever I produce, I have to produce a preset number of both the 4430 and 4460.
So - in that case, if I get a huge order for just the 4430 while the 4460s are not rolling out as quickly as I'd hoped - I'll just continue making wafers that way. If the 4460 part is on 50% of the wafer, I'll just wastefully warehouse them to meet 4430 orders - or - I'll back-order the 4430 until someone ups my 4460 orders.
Not reasonable.
Case 3 - I can just bin them, they're the same part.
Not according to the design documents, they are not that way one bit.
Not reasonable.
~~~~~~~~
As I insisted before - people are trying to project what they think happens based on old PC processor tricks, way out of date for what is going on today.
To me, making 4430s and 4460s on the same wafer sounds exactly like mixing chocolate chip and oatmeal cookie dough in the same bowl at the same time - so, personally, I don't think they're doing it.
But I could be wrong, I often am, and nothing is impossible.
I simply want a real reference from the industry to support this claim, because I find it outrageous.
If these were low-yield, low-demand chips, sure, I'd buy it. But with the SGS2 moving 10+ million units already as a basis for worldwide Gnex projections and who knows how many 4430 products are going to sell in this lifetime, why would I increase my manufacturing costs?
