On Fri, 2020-06-05 at 13:46 +0800, Hongtao Liu via Gcc-patches wrote:
> Hi:
>
> +/* Optimize vector MUL generation for V8QI, V16QI and V32QI
> + under TARGET_AVX512BW. i.e. for v16qi a * b, it has
> +
> + vpmovzxbw ymm2, xmm0
> + vpmovzxbw ymm3, xmm1
> + vpmullw ymm4, ymm2, ymm3
> + vpmovwb xmm0, ymm4
> +
> + it would take less instructions than ix86_expand_vecop_qihi.
> + Return true if success. */
>
> Bootstrap is ok, regression test on i386/x86-64 backend is ok.
>
> gcc/ChangeLog:
> PR target/95488
> * config/i386/i386-expand.c (ix86_expand_vecmul_qihi): New
> function.
> * config/i386/i386-protos.h (ix86_expand_vecmul_qihi): Declare.
> * config/i386/sse.md (mul<mode>3): Drop mask_name since
> there's no real vector char multiplication instruction with
> mask. Also optimize it under TARGET_AVX512BW.
> (mulv8qi3): New expander.
>
> gcc/testsuite/ChangeLog:
> * gcc.target/i386/avx512bw-pr95488-1.c: New test.
> * gcc.target/i386/avx512bw-pr95488-2.c: Ditto.
> * gcc.target/i386/avx512vl-pr95488-1.c: Ditto.
> * gcc.target/i386/avx512vl-pr95488-2.c: Ditto.
>
OK
jeff
