diff options
| author | Jakub Jelinek <jakub@redhat.com> | 2013-04-29 09:43:20 +0200 |
|---|---|---|
| committer | Jakub Jelinek <jakub@gcc.gnu.org> | 2013-04-29 09:43:20 +0200 |
| commit | baee1763e2c82dab41d55da559292030974b41c6 (patch) | |
| tree | 0e4b8e9c518b7c25c2e179d7eba8512edf482d22 /gcc/loop-init.c | |
| parent | 8f36b23705408ae34094fe24b3a6d155a98b5cb7 (diff) | |
| download | gcc-baee1763e2c82dab41d55da559292030974b41c6.zip gcc-baee1763e2c82dab41d55da559292030974b41c6.tar.gz gcc-baee1763e2c82dab41d55da559292030974b41c6.tar.bz2 | |
predicates.md (general_vector_operand): New predicate.
* config/i386/predicates.md (general_vector_operand): New predicate.
* config/i386/i386.c (const_vector_equal_evenodd_p): New function.
(ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
if they aren't nonimmediate operands. If their original values
satisfy const_vector_equal_evenodd_p, don't shift them.
* config/i386/sse.md (mul<mode>3): Use general_vector_operand
predicates. For the SSE4.1 case force operands[{1,2}] into registers
if not nonimmediate_operand.
(vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
instead of register_operand.
(vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
From-SVN: r198387
Diffstat (limited to 'gcc/loop-init.c')
0 files changed, 0 insertions, 0 deletions