-qstrict

In IBM® XL C/C++ for AIX® 16.1.0 or earlier releases, -O3 implies -qnostrict, which enables many floating-point optimizations. However, in IBM Open XL C/C++ for AIX 17.1.3, -O3 does not enable most of the floating-point optimizations that old releases enabled. If -O3 -qstrict was used in classic XL compilers, it maps to just -O3 in Open XL. To enable more floating-point optimizations in IBM Open XL C/C++ for AIX 17.1.3, try the following steps:

1. Specify -O3 to let the program produce expected results.
2. Specify -Ofast to further optimize the program.
3. Customize the usage of the following options to achieve the right balance between accuracy in floating point operations and speed:
   • -f[no-]honor-infinities
   • -f[no-]honor-nans
   • -f[no-]math-errno
   • -f[no-]finite-math-only
   • -f[no-]associative-math
   • -f[no-]reciprocal-math
   • -f[no-]signed-zeros
   • -f[no-]trapping-math
   • -ffp-contract
   • -f[no-]rounding-math