Need vector math functions with non-AOCC compilers and need more functions

[ye-luo]

Found the following message in amdlibm_vec.h

/*
 * We don’t expect anyone to include this file directly.
* but rely on AOCC to generate appropriate calls to vector functions
*/

#ifndef AMD_LIBM_VEC_EXPERIMENTAL
#error "The functionality in this header is experimental, please define AMD_LIBM_VEC_EXPERIMENTAL"
#endif /* EXPERIMENTAL */

But users may use these vector math functions with upstream Clang or GCC.
How risky is it to use with non-AOCC compilers. What is the major risk? Memroy alignment requirements?

Historically, AMD LibM does provide even more funcitons like
vrsa_sincosf and vrda_sincos
Please consider to add them.

[pmallapp-amd]

Hi Ye-Luo,
Thanks for the suggestions, we had to remove them as they were not required by many,
One can create wrapper arround the vrs4/vrs8/vrs16 to achieve the same.