|more useful options|
Re^7: subtraction issueby syphilis (Chancellor)
|on Jan 28, 2017 at 06:59 UTC||Need Help??|
... assuming that mpfr is using the exact value of the double precision ...
It is - unless there's a bug at work.
which makes d = 58
Here's a couple of values that, according to mpfr, require 60 decimal digits:
Might be time I stopped fiddling and started thinking. (Have you got a spare one of those in-skull emulators that you could send over ?)
Update: Fiddling is much easier, especially on a Sat'dy night. I've found numerous doubles that apparently require 122 decimal digits for exact representation, but haven't found any that require more than 122. For example: