Originally posted by lewdvig
There are lots of good reasons to make an Altivec G3. It has better thermal characteristics than the G5, smaller die, fewer transitors, lower cost to manufacture. They will milk this design as long as they can.
Perhaps. But it would take development work to make an Altivec G3, and if that same development work was put into the G5 or G6, they could get the same advantages with a much better processor.
Originally posted by lewdvig
And from what I have read IBM is skipping the 90nm process for the G5. They are going straight to 60.
That was on Page 2 for a reason. In any case, they wouldn't skip to 60 if they wouldn't get it almost as quickly as they could get 90. And a 60 nm G5 (or G6) would be even better for PowerBooks and iBooks.
Originally posted by lewdvig
Apple will need some product differentiation between the Pro and Consumer lineup. G4 (if the next 750 is indeed called a 'G4' by Apple) versus G5.
Apple strives to have one and only one processor platform. They had that with the G3, and they'll get it with the G5 or G6. It's far more simple and effective. The only reason they didn't have it with the G4 was Motorola's constant shortages. With IBM, I think they'll go back to the single processor platform. Makes it easier to have a unified architecture, for sure.
Originally posted by lewdvig
Thermal characteristics are as much a function of good design as process these days. The thermal characteristics of NVidia's NV30 part were horrible despite it being based on a 130nm process (vs 150nm).
IBM is good design. Yes, they could make the G3 into a killer Altivec platform. But they could just as easily make the G5 into the kind of killer chip that can go all across the line.