The following moves the check whether the maximum vectorization
factor determined by data dependence analysis is in conflict with
the chosen vectorization factor to after the point where we applied
both the SLP and the unrolling adjustment to the vectorization
factor. We check the latter before applying unrolling, but the
SLP adjustment can result in both missed optimization and wrong-code.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
* tree-vect-loop.cc (vect_analyze_loop_2): Move check
of VF against max_vf until VF is final.
---
gcc/tree-vect-loop.cc | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index 58679e91c0a..a73a533beb1 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2817,9 +2817,6 @@ vect_analyze_loop_2 (loop_vec_info loop_vinfo, bool
&fatal,
"can't determine vectorization factor.\n");
return ok;
}
- if (max_vf != MAX_VECTORIZATION_FACTOR
- && maybe_lt (max_vf, LOOP_VINFO_VECT_FACTOR (loop_vinfo)))
- return opt_result::failure_at (vect_location, "bad data dependence.\n");
/* Compute the scalar iteration cost. */
vect_compute_single_scalar_iteration_cost (loop_vinfo);
@@ -2881,6 +2878,10 @@ start_over:
LOOP_VINFO_INT_NITERS (loop_vinfo));
}
+ if (max_vf != MAX_VECTORIZATION_FACTOR
+ && maybe_lt (max_vf, LOOP_VINFO_VECT_FACTOR (loop_vinfo)))
+ return opt_result::failure_at (vect_location, "bad data dependence.\n");
+
loop_vinfo->vector_costs = init_cost (loop_vinfo, false);
/* Analyze the alignment of the data-refs in the loop.
--
2.35.3
