diff options
Diffstat (limited to 'gcc/tree-vect-loop.cc')
| -rw-r--r-- | gcc/tree-vect-loop.cc | 36 |
1 files changed, 16 insertions, 20 deletions
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc index 9320bf8..50cdc2a 100644 --- a/gcc/tree-vect-loop.cc +++ b/gcc/tree-vect-loop.cc @@ -738,7 +738,6 @@ _loop_vec_info::_loop_vec_info (class loop *loop_in, vec_info_shared *shared) nonlinear_iv (false), ivexpr_map (NULL), scan_map (NULL), - slp_unrolling_factor (1), inner_loop_cost_factor (param_vect_inner_loop_cost_factor), vectorizable (false), can_use_partial_vectors_p (param_vect_partial_vector_usage != 0), @@ -2236,16 +2235,15 @@ start_over: if (!ok) return ok; - /* If there are any SLP instances mark them as pure_slp. */ + /* If there are any SLP instances mark them as pure_slp and compute + the overall vectorization factor. */ if (!vect_make_slp_decision (loop_vinfo)) return opt_result::failure_at (vect_location, "no stmts to vectorize.\n"); if (dump_enabled_p ()) dump_printf_loc (MSG_NOTE, vect_location, "Loop contains only SLP stmts\n"); - /* Determine the vectorization factor from the SLP decision. */ - LOOP_VINFO_VECT_FACTOR (loop_vinfo) - = LOOP_VINFO_SLP_UNROLLING_FACTOR (loop_vinfo); + /* Dump the vectorization factor from the SLP decision. */ if (dump_enabled_p ()) { dump_printf_loc (MSG_NOTE, vect_location, "vectorization factor = "); @@ -2253,12 +2251,6 @@ start_over: dump_printf (MSG_NOTE, "\n"); } - /* Optimize the SLP graph with the vectorization factor fixed. */ - vect_optimize_slp (loop_vinfo); - - /* Gather the loads reachable from the SLP graph entries. */ - vect_gather_slp_loads (loop_vinfo); - /* We don't expect to have to roll back to anything other than an empty set of rgroups. */ gcc_assert (LOOP_VINFO_MASKS (loop_vinfo).is_empty ()); @@ -2273,6 +2265,12 @@ start_over: poly_uint64 vectorization_factor = LOOP_VINFO_VECT_FACTOR (loop_vinfo); gcc_assert (known_ne (vectorization_factor, 0U)); + /* Optimize the SLP graph with the vectorization factor fixed. */ + vect_optimize_slp (loop_vinfo); + + /* Gather the loads reachable from the SLP graph entries. */ + vect_gather_slp_loads (loop_vinfo); + if (LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo) && dump_enabled_p ()) { dump_printf_loc (MSG_NOTE, vect_location, @@ -2598,7 +2596,7 @@ again: stmt_vec_info vinfo; vinfo = SLP_TREE_SCALAR_STMTS (SLP_INSTANCE_TREE (instance))[0]; - if (! STMT_VINFO_GROUPED_ACCESS (vinfo)) + if (!vinfo || !STMT_VINFO_GROUPED_ACCESS (vinfo)) continue; vinfo = DR_GROUP_FIRST_ELEMENT (vinfo); unsigned int size = DR_GROUP_SIZE (vinfo); @@ -7149,11 +7147,7 @@ vectorizable_reduction (loop_vec_info loop_vinfo, reduction variable. */ slp_tree *slp_op = XALLOCAVEC (slp_tree, op.num_ops); tree *vectype_op = XALLOCAVEC (tree, op.num_ops); - /* We need to skip an extra operand for COND_EXPRs with embedded - comparison. */ - unsigned opno_adjust = 0; - if (op.code == COND_EXPR && COMPARISON_CLASS_P (op.ops[0])) - opno_adjust = 1; + gcc_assert (op.code != COND_EXPR || !COMPARISON_CLASS_P (op.ops[0])); for (i = 0; i < (int) op.num_ops; i++) { /* The condition of COND_EXPR is checked in vectorizable_condition(). */ @@ -7163,7 +7157,7 @@ vectorizable_reduction (loop_vec_info loop_vinfo, stmt_vec_info def_stmt_info; enum vect_def_type dt; if (!vect_is_simple_use (loop_vinfo, slp_for_stmt_info, - i + opno_adjust, &op.ops[i], &slp_op[i], &dt, + i, &op.ops[i], &slp_op[i], &dt, &vectype_op[i], &def_stmt_info)) { if (dump_enabled_p ()) @@ -7174,12 +7168,14 @@ vectorizable_reduction (loop_vec_info loop_vinfo, /* Skip reduction operands, and for an IFN_COND_OP we might hit the reduction operand twice (once as definition, once as else). */ - if (op.ops[i] == op.ops[STMT_VINFO_REDUC_IDX (stmt_info)]) + if (SLP_TREE_CHILDREN (slp_for_stmt_info)[i] + == SLP_TREE_CHILDREN + (slp_for_stmt_info)[SLP_TREE_REDUC_IDX (slp_for_stmt_info)]) continue; /* There should be only one cycle def in the stmt, the one leading to reduc_def. */ - if (VECTORIZABLE_CYCLE_DEF (dt)) + if (SLP_TREE_CHILDREN (slp_for_stmt_info)[i]->cycle_info.id != -1) return false; if (!vectype_op[i]) |