Re: [PATCH V2] [i386] Optimize v4si broadcast for noavx512vl.

2022-03-07 Thread Uros Bizjak via Gcc-patches
On Mon, Mar 7, 2022 at 6:11 AM liuhongt wrote: > > >What happens if you set preferred_for_speed to false for alternative 1? > It works, and I've removed the newly added splitter in this patch. > Also i tried to do similar things to *vec_dup with mode iterator > AVX2_VEC_DUP_MODE, but it hit ICE d

Re: [PATCH V2] [i386] Optimize v4si broadcast for noavx512vl.

2022-03-06 Thread Hongtao Liu via Gcc-patches
Met some problem in git send-email --cc=a,b,c, so manually CC. On Mon, Mar 7, 2022 at 1:11 PM liuhongt via Gcc-patches wrote: > > >What happens if you set preferred_for_speed to false for alternative 1? > It works, and I've removed the newly added splitter in this patch. > Also i tried to do simi

[PATCH V2] [i386] Optimize v4si broadcast for noavx512vl.

2022-03-06 Thread liuhongt via Gcc-patches
>What happens if you set preferred_for_speed to false for alternative 1? It works, and I've removed the newly added splitter in this patch. Also i tried to do similar things to *vec_dup with mode iterator AVX2_VEC_DUP_MODE, but it hit ICE during reload since x86 don't have direct move for QImode