https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120428
Jonathan Wakely <redi at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Known to fail| |14.3.0, 15.1.0
CC| |liuhongt at gcc dot gnu.org
Known to work| |14.2.0
Ever confirmed|0 |1
Component|tree-optimization |target
Status|UNCONFIRMED |NEW
Summary|[15/16 regression] |[14/15/16 regression]
|Suboptimal autovec |Suboptimal autovec
|involving blocked |involving blocked
|permutation and std::copy |permutation and std::copy
Last reconfirmed| |2025-05-27
--- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> ---
The regression was caused by r15-3078-g6ea25c041964bf63014fcf7bb68fb1f5a0a4e123
Align ix86_{move_max,store_max} with vectorizer.
When none of mprefer-vector-width, avx256_optimal/avx128_optimal,
avx256_store_by_pieces/avx512_store_by_pieces is specified, GCC will
set ix86_{move_max,store_max} as max available vector length except
for AVX part.
if (TARGET_AVX512F_P (opts->x_ix86_isa_flags)
&& TARGET_EVEX512_P (opts->x_ix86_isa_flags2))
opts->x_ix86_move_max = PVW_AVX512;
else
opts->x_ix86_move_max = PVW_AVX128;
So for -mavx2, vectorizer will choose 256-bit for vectorization, but
128-bit is used for struct copy, there could be a potential STLF issue
due to this "misalign".
The patch fixes that.
gcc/ChangeLog:
* config/i386/i386-options.cc (ix86_option_override_internal):
set ix86_{move_max,store_max} to PVW_AVX256 when TARGET_AVX
instead of PVW_AVX128.
