I am looking for documentation that clarifies which IEEE 754 rounding mode the Visual Studio (2022) C++ compiler uses when the _RC_NEAR
flag is used to switch floating-point rounding modes.
This MS doc page shows examples for switching rounding to + and - infinity, along with specifying the default rounding method (turns out it is _RC_NEAR
if you look in float.h
), but I can't find any more detail about _RC_NEAR
, if that corresponds to ties to even or ties away from zero, as mentioned on the wikipedia page.