Bruce Momjian <pgman@candle.pha.pa.us> writes:
> I will run it some more tomorrow but clearly we are
> seeing reasonable numbers now.
... which still have no provable relationship to the ratio we need to
measure. See my previous comments to Curt; I don't think you can
possibly get trustworthy results out of a shell script + dd approach,
because we do not implement Postgres using dd.
If you implemented a C testbed and then proved by experiment that the
shell script got comparable numbers, then I'd believe its results.
Without that confirmation, these are just meaningless numbers.
regards, tom lane