The slide is talking about RAMP comparisons. That strongly implies that time also increases as you move from left to right, since any decrease in volume indicates the end of the ramp, so data from that point on is probably not shown.
I'll buy that. I was hoping AMD might have inadvertantly released evidence of demand-constraint. *sigh*
Though as I think Alan pointed out, the use of volume instead of time makes the 130nm SOI line much steeper and shorter than it would otherwise be. All in all, yield improvements probably have occurred on par with the other processes, which is good news I'd say. Now, I don't know a damn thing about it, but I wouldn't think that huge volumes would make it any easier to get better yields, whereas lots of time should.