There's a discrepancy in SLP vs non-SLP vectorization that SLP build
does not handle plain SSA copies (which should have been elimiated
earlier). But this now bites back since non-SLP happily handles them,
causing a regression with --param vect-force-slp=1 which is now default,
resulting in a big performance regression in 456.hmmer.
So the following restores parity between SLP and non-SLP here, defering
the missed copy elimination to later (PR118565).
Bootstrap and regtest running on x86_64-unknown-linux-gnu.
PR tree-optimization/117875
* tree-vect-slp.cc (vect_build_slp_tree_1): Handle SSA copies.
---
gcc/tree-vect-slp.cc | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
index 1d8e62e2fc1..5d9a206f8f5 100644
--- a/gcc/tree-vect-slp.cc
+++ b/gcc/tree-vect-slp.cc
@@ -1522,7 +1522,8 @@ vect_build_slp_tree_1 (vec_info *vinfo, unsigned char
*swap,
&& TREE_CODE_CLASS (tree_code (rhs_code)) != tcc_comparison
&& rhs_code != VIEW_CONVERT_EXPR
&& rhs_code != CALL_EXPR
- && rhs_code != BIT_FIELD_REF)
+ && rhs_code != BIT_FIELD_REF
+ && rhs_code != SSA_NAME)
{
if (dump_enabled_p ())
dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
--
2.43.0
