[Bug tree-optimization/123343] Loop unrolling before vectorization produces suboptimal RISC-V code

[chenzhongyao.hit at gmail dot com via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123343

--- Comment #9 from Zhongyao Chen <chenzhongyao.hit at gmail dot com> ---
New find, seems that multi-reduction SLP group formation is the cause.

if I skip the SLP group with:

diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
index 70470612411..425842e90d9 100644
--- a/gcc/tree-vect-slp.cc
+++ b/gcc/tree-vect-slp.cc
@@ -4834,7 +4834,7 @@ vect_analyze_slp_reductions (loop_vec_info loop_vinfo,
        }
     }

-  if (scalar_stmts.length () > 1)
+  if (0 && scalar_stmts.length () > 1)
     {
       /* Try to form a reduction group.  */
       unsigned int group_size = scalar_stmts.length ();

it produces the expected efficient assembly. So we may need a cost check to
reject multi-reduction SLP groups here.