On Thu, Nov 24, 2022 at 4:53 PM Jakub Jelinek wrote:
>
> On Thu, Nov 24, 2022 at 09:22:00AM +0800, liuhongt via Gcc-patches wrote:
> > --- a/gcc/config/i386/i386.md
> > +++ b/gcc/config/i386/i386.md
> > @@ -130,6 +130,7 @@ (define_c_enum "unspec" [
> >;; For AVX/AVX512F support
> >UNSPEC_S
On Thu, Nov 24, 2022 at 09:22:00AM +0800, liuhongt via Gcc-patches wrote:
> --- a/gcc/config/i386/i386.md
> +++ b/gcc/config/i386/i386.md
> @@ -130,6 +130,7 @@ (define_c_enum "unspec" [
>;; For AVX/AVX512F support
>UNSPEC_SCALEF
>UNSPEC_PCMP
> + UNSPEC_CVTBFSF
>
>;; Generic math
After supporting real __bf16, the implementation of _mm_cvtsbh_ss went
wrong.
The patch add a builtin to generate pslld for the intrinsic, also
extendbfsf2 is supported with pslld when !flag_signaling_nans &&
!HONOR_NANS (BFmode).
truncsfbf2 is supported with vcvtneps2bf16 when !flag_signaling_na