128 bit integers in ifort

[Stefanos_K_]

I recently became aware that ifort now supports 128-bit reals via the -r16 flag. Why only reals? Why not introduce a -i16 flag to support 128-bit integers? Is there any chance this feature will be incorporated into ifort?

[Steven_L_Intel1]

128-bit reals, with extended precision and range, have been supported by Intel Fortran since 2004, and DEC Fortran since 1980. The preferable way to use these is to declare variables as REAL(16) - use of flags to change default kinds is not recommended. We do not have current plans for 128-bit integer support.

[TimP]

Full 128-bit integer support would require development of a full library to support corresponding operations.

gnu compilers offer (in gnu modes, but not under standard compliance flags, and not on all CPU targets) 128-bit integers without supporting arithmetic operations.  Those are simply alternatives to what can be done with __m128i data type in Intel C++ as well.  Note that Microsoft has such data types only for IA-compatible CPUs.