Intel Corp (INTC)

[Elmer Phud]

I thought that it was fairly common knowledge that for a design change, there will be a significant amount initial and on-going "tweaking" necessary to improve the process. As such, yields will generally improve as the process is improved, but may be poor initially. Am I incorrect in this assumption?

Yes you may be incorrect. The defect density for a process doesn't know you just did a stepping change. Most stepping changes are metal only in which case the effect on yields would not be noticeable unless AMD is scrapping die on the wafer for speed and that is not very reliable. For every full mask layer stepping change, there are usually a number of additional metal only steppings so it should inherit the same yield as it's predecessor.