On Mon, 12 Jun 2017, Matthew Fortune wrote: > The only benefit would be if something could fill the load latency between > obtaining the offset and adding to the base which seems low probability.
Or if any intermediate results were found to be reusable in other calculations nearby (not that I think this is likely though). > Thanks for the intricate commentary on this patch, makes it very easy to > review. I hoped it would. Writing a proper description also serves as the last point of reflection: "Did I get it right? Haven't I missed anything?" Patch applied now, thank you for you review. Maciej
