diff options
| author | jakub <jakub@138bc75d-0d04-0410-961f-82ee72b054a4> | 2013-10-31 13:57:05 +0000 |
|---|---|---|
| committer | jakub <jakub@138bc75d-0d04-0410-961f-82ee72b054a4> | 2013-10-31 13:57:05 +0000 |
| commit | c8a2b4ff9b521b6b455caf9b45137c7153738a64 (patch) | |
| tree | 724d536345f0d153b316b5c50064d6336fea231b /gcc/tree-vect-loop.c | |
| parent | fc08b993096e47f63278b70ea0df4e3ae54115cc (diff) | |
| download | gcc-c8a2b4ff9b521b6b455caf9b45137c7153738a64.tar.gz | |
* tree.c (tree_ctz): New function.
* tree.h (tree_ctz): New prototype.
* tree-ssanames.h (get_range_info, get_nonzero_bits): Change
first argument from tree to const_tree.
* tree-ssanames.c (get_range_info, get_nonzero_bits): Likewise.
* tree-vectorizer.h (vect_generate_tmps_on_preheader): New prototype.
* tree-vect-loop-manip.c (vect_generate_tmps_on_preheader): No longer
static.
* expr.c (highest_pow2_factor): Reimplemented using tree_ctz.
* tree-vect-loop.c (vect_analyze_loop_operations,
vect_transform_loop): Don't force scalar loop for bound just because
number of iterations is unknown, only do it if it is not known to be
a multiple of vectorization_factor.
* builtins.c (get_object_alignment_2): Use tree_ctz on offset.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@204257 138bc75d-0d04-0410-961f-82ee72b054a4
Diffstat (limited to 'gcc/tree-vect-loop.c')
| -rw-r--r-- | gcc/tree-vect-loop.c | 21 |
1 files changed, 13 insertions, 8 deletions
diff --git a/gcc/tree-vect-loop.c b/gcc/tree-vect-loop.c index 296b7b24cc4..1da87c7201e 100644 --- a/gcc/tree-vect-loop.c +++ b/gcc/tree-vect-loop.c @@ -1586,9 +1586,9 @@ vect_analyze_loop_operations (loop_vec_info loop_vinfo, bool slp) return false; } - if (!LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo) - || LOOP_VINFO_INT_NITERS (loop_vinfo) % vectorization_factor != 0 - || LOOP_PEELING_FOR_ALIGNMENT (loop_vinfo)) + if (LOOP_PEELING_FOR_ALIGNMENT (loop_vinfo) + || ((int) tree_ctz (LOOP_VINFO_NITERS (loop_vinfo)) + < exact_log2 (vectorization_factor))) { if (dump_enabled_p ()) dump_printf_loc (MSG_NOTE, vect_location, "epilog loop required.\n"); @@ -5656,15 +5656,20 @@ vect_transform_loop (loop_vec_info loop_vinfo) will remain scalar and will compute the remaining (n%VF) iterations. (VF is the vectorization factor). */ - if (!LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo) - || (LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo) - && LOOP_VINFO_INT_NITERS (loop_vinfo) % vectorization_factor != 0) - || LOOP_VINFO_PEELING_FOR_GAPS (loop_vinfo)) + if ((int) tree_ctz (LOOP_VINFO_NITERS (loop_vinfo)) + < exact_log2 (vectorization_factor) + || LOOP_VINFO_PEELING_FOR_GAPS (loop_vinfo)) vect_do_peeling_for_loop_bound (loop_vinfo, &ratio, th, check_profitability); - else + else if (LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo)) ratio = build_int_cst (TREE_TYPE (LOOP_VINFO_NITERS (loop_vinfo)), LOOP_VINFO_INT_NITERS (loop_vinfo) / vectorization_factor); + else + { + tree ni_name, ratio_mult_vf; + vect_generate_tmps_on_preheader (loop_vinfo, &ni_name, &ratio_mult_vf, + &ratio, NULL); + } /* 1) Make sure the loop header has exactly two entries 2) Make sure we have a preheader basic block. */ |