This splits out load permutation computation from SLP tree building.
It also removes the broken support for swapping mismatched operands.
If it ever triggers we'll ICE later because:
case vect_internal_def:
! if (different_types)
! {
! oprnd0_info = (*oprnds_info)[0];
! oprnd1_info = (*oprnds_info)[0];
! if (i == 0)
! oprnd1_info->def_stmts.quick_push (def_stmt);
! else
! oprnd0_info->def_stmts.quick_push (def_stmt);
! }
pushes to the same operand vector twice ... The cases this
tries to handle should all be canonicalized by reassoc earlier.
I am going to re-instantiate more complete support for handling
commutated operations in the next patch.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2013-04-10 Richard Biener <rguent...@suse.de>
* tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
broken code swapping operands.
(vect_build_slp_tree): Do not compute load permutations here.
(vect_analyze_slp_instance): Compute load permutations here,
after building the SLP tree.
Index: gcc/tree-vect-slp.c
===================================================================
*** gcc/tree-vect-slp.c (revision 197635)
--- gcc/tree-vect-slp.c (working copy)
*************** vect_get_and_check_slp_defs (loop_vec_in
*** 204,218 ****
{
tree oprnd;
unsigned int i, number_of_oprnds;
! tree def, def_op0 = NULL_TREE;
gimple def_stmt;
enum vect_def_type dt = vect_uninitialized_def;
- enum vect_def_type dt_op0 = vect_uninitialized_def;
struct loop *loop = NULL;
- enum tree_code rhs_code;
- bool different_types = false;
bool pattern = false;
! slp_oprnd_info oprnd_info, oprnd0_info, oprnd1_info;
int op_idx = 1;
tree compare_rhs = NULL_TREE;
--- 204,215 ----
{
tree oprnd;
unsigned int i, number_of_oprnds;
! tree def;
gimple def_stmt;
enum vect_def_type dt = vect_uninitialized_def;
struct loop *loop = NULL;
bool pattern = false;
! slp_oprnd_info oprnd_info;
int op_idx = 1;
tree compare_rhs = NULL_TREE;
*************** vect_get_and_check_slp_defs (loop_vec_in
*** 334,345 ****
oprnd_info->first_def_type = NULL_TREE;
oprnd_info->first_const_oprnd = oprnd;
}
-
- if (i == 0)
- {
- def_op0 = def;
- dt_op0 = dt;
- }
}
else
{
--- 331,336 ----
*************** vect_get_and_check_slp_defs (loop_vec_in
*** 357,413 ****
TREE_TYPE (def))))
|| (!def
&& !types_compatible_p (TREE_TYPE
(oprnd_info->first_const_oprnd),
! TREE_TYPE (oprnd)))
! || different_types)
{
! if (number_of_oprnds != 2)
! {
! if (dump_enabled_p ())
! dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
! "Build SLP failed: different types ");
!
! return false;
! }
!
! /* Try to swap operands in case of binary operation. */
! if (i == 0)
! different_types = true;
! else
! {
! oprnd0_info = (*oprnds_info)[0];
! if (is_gimple_assign (stmt)
! && (rhs_code = gimple_assign_rhs_code (stmt))
! && TREE_CODE_CLASS (rhs_code) == tcc_binary
! && commutative_tree_code (rhs_code)
! && oprnd0_info->first_dt == dt
! && oprnd_info->first_dt == dt_op0
! && def_op0 && def
! && !(oprnd0_info->first_def_type
! && !types_compatible_p (oprnd0_info->first_def_type,
! TREE_TYPE (def)))
! && !(oprnd_info->first_def_type
! && !types_compatible_p (oprnd_info->first_def_type,
! TREE_TYPE (def_op0))))
! {
! if (dump_enabled_p ())
! {
! dump_printf_loc (MSG_NOTE, vect_location,
! "Swapping operands of ");
! dump_gimple_stmt (MSG_NOTE, TDF_SLIM, stmt, 0);
! }
!
! swap_tree_operands (stmt, gimple_assign_rhs1_ptr (stmt),
! gimple_assign_rhs2_ptr (stmt));
! }
! else
! {
! if (dump_enabled_p ())
! dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
! "Build SLP failed: different types ");
! return false;
! }
! }
}
}
--- 348,360 ----
TREE_TYPE (def))))
|| (!def
&& !types_compatible_p (TREE_TYPE
(oprnd_info->first_const_oprnd),
! TREE_TYPE (oprnd))))
{
! if (dump_enabled_p ())
! dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
! "Build SLP failed: different types ");
! return false;
}
}
*************** vect_get_and_check_slp_defs (loop_vec_in
*** 420,437 ****
break;
case vect_internal_def:
! if (different_types)
! {
! oprnd0_info = (*oprnds_info)[0];
! oprnd1_info = (*oprnds_info)[0];
! if (i == 0)
! oprnd1_info->def_stmts.quick_push (def_stmt);
! else
! oprnd0_info->def_stmts.quick_push (def_stmt);
! }
! else
! oprnd_info->def_stmts.quick_push (def_stmt);
!
break;
default:
--- 367,373 ----
break;
case vect_internal_def:
! oprnd_info->def_stmts.quick_push (def_stmt);
break;
default:
*************** static bool
*** 460,468 ****
vect_build_slp_tree (loop_vec_info loop_vinfo, bb_vec_info bb_vinfo,
slp_tree *node, unsigned int group_size,
unsigned int *max_nunits,
- vec<int> *load_permutation,
vec<slp_tree> *loads,
! unsigned int vectorization_factor, bool *loads_permuted)
{
unsigned int i;
vec<gimple> stmts = SLP_TREE_SCALAR_STMTS (*node);
--- 396,403 ----
vect_build_slp_tree (loop_vec_info loop_vinfo, bb_vec_info bb_vinfo,
slp_tree *node, unsigned int group_size,
unsigned int *max_nunits,
vec<slp_tree> *loads,
! unsigned int vectorization_factor)
{
unsigned int i;
vec<gimple> stmts = SLP_TREE_SCALAR_STMTS (*node);
*************** vect_build_slp_tree (loop_vec_info loop_
*** 478,485 ****
enum machine_mode vec_mode;
struct data_reference *first_dr;
HOST_WIDE_INT dummy;
- bool permutation = false;
- unsigned int load_place;
gimple first_load = NULL, prev_first_load = NULL, old_first_load = NULL;
vec<slp_oprnd_info> oprnds_info;
unsigned int nops;
--- 413,418 ----
*************** vect_build_slp_tree (loop_vec_info loop_
*** 839,854 ****
}
}
- /* Store the place of this load in the interleaving chain. In
- case that permutation is needed we later decide if a specific
- permutation is supported. */
- load_place = vect_get_place_in_interleaving_chain (stmt,
- first_load);
- if (load_place != i)
- permutation = true;
-
- load_permutation->safe_push (load_place);
-
/* We stop the tree when we reach a group of loads. */
stop_recursion = true;
continue;
--- 772,777 ----
*************** vect_build_slp_tree (loop_vec_info loop_
*** 925,941 ****
if (stop_recursion)
{
loads->safe_push (*node);
- if (permutation)
- *loads_permuted = true;
- else
- {
- /* We don't check here complex numbers chains, so we set
- LOADS_PERMUTED for further check in
- vect_supported_load_permutation_p. */
- if (rhs_code == REALPART_EXPR || rhs_code == IMAGPART_EXPR)
- *loads_permuted = true;
- }
-
vect_free_oprnd_info (oprnds_info);
return true;
}
--- 848,853 ----
*************** vect_build_slp_tree (loop_vec_info loop_
*** 951,958 ****
child = vect_create_new_slp_node (oprnd_info->def_stmts);
if (!child
|| !vect_build_slp_tree (loop_vinfo, bb_vinfo, &child, group_size,
! max_nunits, load_permutation, loads,
! vectorization_factor, loads_permuted))
{
if (child)
oprnd_info->def_stmts = vNULL;
--- 863,870 ----
child = vect_create_new_slp_node (oprnd_info->def_stmts);
if (!child
|| !vect_build_slp_tree (loop_vinfo, bb_vinfo, &child, group_size,
! max_nunits, loads,
! vectorization_factor))
{
if (child)
oprnd_info->def_stmts = vNULL;
*************** vect_analyze_slp_instance (loop_vec_info
*** 1604,1613 ****
unsigned int vectorization_factor = 0;
int i;
unsigned int max_nunits = 0;
- vec<int> load_permutation;
vec<slp_tree> loads;
struct data_reference *dr = STMT_VINFO_DATA_REF (vinfo_for_stmt (stmt));
- bool loads_permuted = false;
vec<gimple> scalar_stmts;
if (GROUP_FIRST_ELEMENT (vinfo_for_stmt (stmt)))
--- 1516,1523 ----
*************** vect_analyze_slp_instance (loop_vec_info
*** 1689,1701 ****
node = vect_create_new_slp_node (scalar_stmts);
- load_permutation.create (group_size * group_size);
loads.create (group_size);
/* Build the tree for the SLP instance. */
if (vect_build_slp_tree (loop_vinfo, bb_vinfo, &node, group_size,
! &max_nunits, &load_permutation, &loads,
! vectorization_factor, &loads_permuted))
{
/* Calculate the unrolling factor based on the smallest type. */
if (max_nunits > nunits)
--- 1599,1610 ----
node = vect_create_new_slp_node (scalar_stmts);
loads.create (group_size);
/* Build the tree for the SLP instance. */
if (vect_build_slp_tree (loop_vinfo, bb_vinfo, &node, group_size,
! &max_nunits, &loads,
! vectorization_factor))
{
/* Calculate the unrolling factor based on the smallest type. */
if (max_nunits > nunits)
*************** vect_analyze_slp_instance (loop_vec_info
*** 1709,1715 ****
"Build SLP failed: unrolling required in basic"
" block SLP");
vect_free_slp_tree (node);
- load_permutation.release ();
loads.release ();
return false;
}
--- 1618,1623 ----
*************** vect_analyze_slp_instance (loop_vec_info
*** 1722,1731 ****
SLP_INSTANCE_BODY_COST_VEC (new_instance) = vNULL;
SLP_INSTANCE_LOADS (new_instance) = loads;
SLP_INSTANCE_FIRST_LOAD_STMT (new_instance) = NULL;
! SLP_INSTANCE_LOAD_PERMUTATION (new_instance) = load_permutation;
if (loads_permuted)
{
if (!vect_supported_load_permutation_p (new_instance, group_size,
load_permutation))
{
--- 1630,1666 ----
SLP_INSTANCE_BODY_COST_VEC (new_instance) = vNULL;
SLP_INSTANCE_LOADS (new_instance) = loads;
SLP_INSTANCE_FIRST_LOAD_STMT (new_instance) = NULL;
! SLP_INSTANCE_LOAD_PERMUTATION (new_instance) = vNULL;
!
! /* Compute the load permutation. */
! slp_tree load_node;
! bool loads_permuted = false;
! vec<int> load_permutation;
! load_permutation.create (group_size * group_size);
! FOR_EACH_VEC_ELT (loads, i, load_node)
! {
! int j;
! gimple load;
! FOR_EACH_VEC_ELT (SLP_TREE_SCALAR_STMTS (load_node), j, load)
! {
! int load_place;
! load_place = vect_get_place_in_interleaving_chain
! (load, GROUP_FIRST_ELEMENT (vinfo_for_stmt (load)));
! if (load_place != j
! /* ??? We allow loads from different groups to
! get to here for a special case handled in
! the permutation code. Make sure we get to that. */
! || (GROUP_FIRST_ELEMENT
! (vinfo_for_stmt (SLP_TREE_SCALAR_STMTS (load_node)[0]))
! != GROUP_FIRST_ELEMENT (vinfo_for_stmt (load))))
! loads_permuted = true;
! load_permutation.safe_push (load_place);
! }
! }
if (loads_permuted)
{
+ SLP_INSTANCE_LOAD_PERMUTATION (new_instance) = load_permutation;
if (!vect_supported_load_permutation_p (new_instance, group_size,
load_permutation))
{
*************** vect_analyze_slp_instance (loop_vec_info
*** 1745,1751 ****
= vect_find_first_load_in_slp_instance (new_instance);
}
else
! SLP_INSTANCE_LOAD_PERMUTATION (new_instance).release ();
/* Compute the costs of this SLP instance. */
vect_analyze_slp_cost (loop_vinfo, bb_vinfo,
--- 1680,1686 ----
= vect_find_first_load_in_slp_instance (new_instance);
}
else
! load_permutation.release ();
/* Compute the costs of this SLP instance. */
vect_analyze_slp_cost (loop_vinfo, bb_vinfo,
*************** vect_analyze_slp_instance (loop_vec_info
*** 1765,1771 ****
/* Failed to SLP. */
/* Free the allocated memory. */
vect_free_slp_tree (node);
- load_permutation.release ();
loads.release ();
return false;
--- 1700,1705 ----
