PowerPC rightfully died, no thanks.
We can't rewrite history, but sure another console contract or two could have prolongated its agony
That said, I'd be curious if they'd have done Xenon II with OOOE back in.
I'm not sure it'd be worth going with such a stepping stone as Redwood though, or at least I'm not sure what the point would be as opposed to just superclocking Xenos and be done with the R&D.
In 2010 there was no good reasonably cheap CPUs for console manufacturers to use outside of custom solutions.
For Xenon, use EDRAM for the L2, increase the L1 size and then the integrated memory controller. With a relatively tiny R&D budget I would have pass at any significant changes, it would have remained an narrow IO core supporting 2 ways SMT. I would not touch the VMX128 units either, improvement or fixs, but same ISA no widening.
Looking at the WiiU CPU size, it could have ended delivering really interesting performances within a tiny package and well at this time there was no other options.
I believe both the Cell and Xenon could have seen one iteration had the manufacturers decided on a shorter hardware cycle. The cell would have been particular as it was so powerful to begin with. A second PPC along with 2MB of shared L2 would already go a hell of a long way balancing things out, then supported for DDR3. Not too crazy on the R&D side of thing and that would have been a hell of a tiny monster.
For the GPU I search for something cheap, not requiring R&D and it sorted of worked with the silicon budget I had in mind which ~170 sq.mm ~Valhalla. Redwood is 104 sq.mm. It would not such crazy thing to do, the XBox and the PS3 used close to off the shelves part for a GPU. Another reason I thought of it instead of simply a bigger Xenos (and a bigger daughter), is that for the low cost I wanted to reach having three chips sounded like a bad idea. So in my mental world, I shifted all the R&D to the CPU and ganging the design from a UMA to a NUMA. In my mental I also chose narrow buses in case they deem it worthy to shrink the chips.
The thing I like with that cheap approach is that the competitor caught its pants would have to wait till 2012 for 28nm to become available and to offer something significantly better and still with no great CPU options arounds, especially CPU that could end in APU.