[Bug target/68923] SSE/AVX movq load (_mm_cvtsi64_si128) not being folded into pmovzx
jakub at gcc dot gnu.org
gcc-bugzilla@gcc.gnu.org
Wed Dec 16 08:53:00 GMT 2015
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68923
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jakub at gcc dot gnu.org,
| |uros at gcc dot gnu.org
--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
To match the -O3 -march=haswell code in the combiner we'd need a pattern
matching something like:
(set (reg:V8SI 94)
(zero_extend:V8SI (vec_select:V8QI (subreg:V16QI (vec_concat:V2DI (mem:DI
(reg/v/f:DI 93 [ p ]) [4 MEM[(uint64_t *)p_2(D)]+0 S8 A64])
(const_int 0 [0])) 0)
(parallel [
(const_int 0 [0])
(const_int 1 [0x1])
(const_int 2 [0x2])
(const_int 3 [0x3])
(const_int 4 [0x4])
(const_int 5 [0x5])
(const_int 6 [0x6])
(const_int 7 [0x7])
]))))
But e.g. for -O2 -mavx2 it is instead:
(set (reg:V8SI 94)
(zero_extend:V8SI (vec_select:V8QI (subreg:V16QI (vec_merge:V2DI
(vec_duplicate:V2DI (mem:DI (reg/v/f:DI 93 [ p ]) [4 MEM[(uint64_t *)p_2(D)]+0
S8 A64]))
(const_vector:V2DI [
(const_int 0 [0])
(const_int 0 [0])
])
(const_int 1 [0x1])) 0)
(parallel [
(const_int 0 [0])
(const_int 1 [0x1])
(const_int 2 [0x2])
(const_int 3 [0x3])
(const_int 4 [0x4])
(const_int 5 [0x5])
(const_int 6 [0x6])
(const_int 7 [0x7])
]))))
More information about the Gcc-bugs
mailing list