https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108040
--- Comment #1 from Martin Jambor <jamborm at gcc dot gnu.org> --- It is probably me not being able to build the necessary cross compiler properly, but I cannot build the provided testcase, I always get errors like the following and then some more: <built-in>: note: initializing argument 2 of ‘__vector(4) float __builtin_ia32_dpbf16ps_v4sf(__vector(4) float, __vector(8) __bf16, __vector(8) __bf16)’ C:/msys64/ucrt64/lib/gcc/x86_64-w64-mingw32/12.2.0/include/avx512bf16vlintrin.h: In function ‘__m128 _mm_mask_dpbf16_ps(__m128, __mmask8, __m128bh, __m128bh)’: C:/msys64/ucrt64/lib/gcc/x86_64-w64-mingw32/12.2.0/include/avx512bf16vlintrin.h:169:57: error: cannot convert ‘__m128bh’ to ‘__vector(8) __bf16’ It would be great if you could simplify the testcase (and also the necessary command line) somewhat - and ideally reproduce it with pristine GCC. Otherwise you may want to report the issue to https://github.com/msys2/MINGW-packages/issues
