[PATCH][GCC 9] AArch64: Add prefer_advsimd_autovec internal tune_flag

[Kyrylo Tkachov via Gcc-patches]

Hi all,

I'd like to add a prefer_advsimd_autovec internal tune_flag that makes GCC pick 
Advanced SIMD over SVE for autovectorisation.
No CPU tuning uses it yet, but I'd like to add this to the GCC 8 and 9 branches 
only as SVE autovectorisation is less mature there and CPUs
may want to prefer Advanced SIMD over SVE when tuning for performance.
This patch provides a minimally invasive way of achieving that.

Bootstrapped and tested on aarch64-none-linux-gnu.
Pushing on the GCC 9 branch and an equivalent GCC 8 branch patch too.

gcc/
        * config/aarch64/aarch64-tuning-flags.def (PREFER_ADVSIMD_AUTOVEC):
        Define.
        * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Use it.
        (aarch64_autovectorize_vector_sizes): Likewise.

prefer-asimd-9.patch
Description: prefer-asimd-9.patch