Feature #904
SmallFpDouble: which impl of InvMod to use?
Start date:
04 Jul 2016
Due date:
% Done:
0%
Estimated time:
Description
The source code for SmallFpDoubleImpl
includes its own impl of InvMod
which uses double
rather than an integral type; there is also a comment saying that the version using double
is faster on some platforms.
Ideally the code should choose which ever impl is faster for the platform.
Can we do this without too much complication?