blob: 310165084a34d1537e6a44a0432299433b5366ec [file] [log] [blame]
/* Analysis Utilities for Loop Vectorization.
Copyright (C) 2006-2020 Free Software Foundation, Inc.
Contributed by Dorit Nuzman <dorit@il.ibm.com>
This file is part of GCC.
GCC is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation; either version 3, or (at your option) any later
version.
GCC is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License
along with GCC; see the file COPYING3. If not see
<http://www.gnu.org/licenses/>. */
#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "backend.h"
#include "rtl.h"
#include "tree.h"
#include "gimple.h"
#include "ssa.h"
#include "expmed.h"
#include "optabs-tree.h"
#include "insn-config.h"
#include "recog.h" /* FIXME: for insn_data */
#include "fold-const.h"
#include "stor-layout.h"
#include "tree-eh.h"
#include "gimplify.h"
#include "gimple-iterator.h"
#include "cfgloop.h"
#include "tree-vectorizer.h"
#include "dumpfile.h"
#include "builtins.h"
#include "internal-fn.h"
#include "case-cfn-macros.h"
#include "fold-const-call.h"
#include "attribs.h"
#include "cgraph.h"
#include "omp-simd-clone.h"
#include "predict.h"
#include "tree-vector-builder.h"
#include "vec-perm-indices.h"
/* Return true if we have a useful VR_RANGE range for VAR, storing it
in *MIN_VALUE and *MAX_VALUE if so. Note the range in the dump files. */
static bool
vect_get_range_info (tree var, wide_int *min_value, wide_int *max_value)
{
value_range_kind vr_type = get_range_info (var, min_value, max_value);
wide_int nonzero = get_nonzero_bits (var);
signop sgn = TYPE_SIGN (TREE_TYPE (var));
if (intersect_range_with_nonzero_bits (vr_type, min_value, max_value,
nonzero, sgn) == VR_RANGE)
{
if (dump_enabled_p ())
{
dump_generic_expr_loc (MSG_NOTE, vect_location, TDF_SLIM, var);
dump_printf (MSG_NOTE, " has range [");
dump_hex (MSG_NOTE, *min_value);
dump_printf (MSG_NOTE, ", ");
dump_hex (MSG_NOTE, *max_value);
dump_printf (MSG_NOTE, "]\n");
}
return true;
}
else
{
if (dump_enabled_p ())
{
dump_generic_expr_loc (MSG_NOTE, vect_location, TDF_SLIM, var);
dump_printf (MSG_NOTE, " has no range info\n");
}
return false;
}
}
/* Report that we've found an instance of pattern PATTERN in
statement STMT. */
static void
vect_pattern_detected (const char *name, gimple *stmt)
{
if (dump_enabled_p ())
dump_printf_loc (MSG_NOTE, vect_location, "%s: detected: %G", name, stmt);
}
/* Associate pattern statement PATTERN_STMT with ORIG_STMT_INFO and
return the pattern statement's stmt_vec_info. Set its vector type to
VECTYPE if it doesn't have one already. */
static stmt_vec_info
vect_init_pattern_stmt (gimple *pattern_stmt, stmt_vec_info orig_stmt_info,
tree vectype)
{
vec_info *vinfo = orig_stmt_info->vinfo;
stmt_vec_info pattern_stmt_info = vinfo->lookup_stmt (pattern_stmt);
if (pattern_stmt_info == NULL)
pattern_stmt_info = orig_stmt_info->vinfo->add_stmt (pattern_stmt);
gimple_set_bb (pattern_stmt, gimple_bb (orig_stmt_info->stmt));
pattern_stmt_info->pattern_stmt_p = true;
STMT_VINFO_RELATED_STMT (pattern_stmt_info) = orig_stmt_info;
STMT_VINFO_DEF_TYPE (pattern_stmt_info)
= STMT_VINFO_DEF_TYPE (orig_stmt_info);
if (!STMT_VINFO_VECTYPE (pattern_stmt_info))
{
gcc_assert (VECTOR_BOOLEAN_TYPE_P (vectype)
== vect_use_mask_type_p (orig_stmt_info));
STMT_VINFO_VECTYPE (pattern_stmt_info) = vectype;
pattern_stmt_info->mask_precision = orig_stmt_info->mask_precision;
}
return pattern_stmt_info;
}
/* Set the pattern statement of ORIG_STMT_INFO to PATTERN_STMT.
Also set the vector type of PATTERN_STMT to VECTYPE, if it doesn't
have one already. */
static void
vect_set_pattern_stmt (gimple *pattern_stmt, stmt_vec_info orig_stmt_info,
tree vectype)
{
STMT_VINFO_IN_PATTERN_P (orig_stmt_info) = true;
STMT_VINFO_RELATED_STMT (orig_stmt_info)
= vect_init_pattern_stmt (pattern_stmt, orig_stmt_info, vectype);
}
/* Add NEW_STMT to STMT_INFO's pattern definition statements. If VECTYPE
is nonnull, record that NEW_STMT's vector type is VECTYPE, which might
be different from the vector type of the final pattern statement.
If VECTYPE is a mask type, SCALAR_TYPE_FOR_MASK is the scalar type
from which it was derived. */
static inline void
append_pattern_def_seq (stmt_vec_info stmt_info, gimple *new_stmt,
tree vectype = NULL_TREE,
tree scalar_type_for_mask = NULL_TREE)
{
gcc_assert (!scalar_type_for_mask
== (!vectype || !VECTOR_BOOLEAN_TYPE_P (vectype)));
vec_info *vinfo = stmt_info->vinfo;
if (vectype)
{
stmt_vec_info new_stmt_info = vinfo->add_stmt (new_stmt);
STMT_VINFO_VECTYPE (new_stmt_info) = vectype;
if (scalar_type_for_mask)
new_stmt_info->mask_precision
= GET_MODE_BITSIZE (SCALAR_TYPE_MODE (scalar_type_for_mask));
}
gimple_seq_add_stmt_without_update (&STMT_VINFO_PATTERN_DEF_SEQ (stmt_info),
new_stmt);
}
/* The caller wants to perform new operations on vect_external variable
VAR, so that the result of the operations would also be vect_external.
Return the edge on which the operations can be performed, if one exists.
Return null if the operations should instead be treated as part of
the pattern that needs them. */
static edge
vect_get_external_def_edge (vec_info *vinfo, tree var)
{
edge e = NULL;
if (loop_vec_info loop_vinfo = dyn_cast <loop_vec_info> (vinfo))
{
e = loop_preheader_edge (loop_vinfo->loop);
if (!SSA_NAME_IS_DEFAULT_DEF (var))
{
basic_block bb = gimple_bb (SSA_NAME_DEF_STMT (var));
if (bb == NULL
|| !dominated_by_p (CDI_DOMINATORS, e->dest, bb))
e = NULL;
}
}
return e;
}
/* Return true if the target supports a vector version of CODE,
where CODE is known to map to a direct optab. ITYPE specifies
the type of (some of) the scalar inputs and OTYPE specifies the
type of the scalar result.
If CODE allows the inputs and outputs to have different type
(such as for WIDEN_SUM_EXPR), it is the input mode rather
than the output mode that determines the appropriate target pattern.
Operand 0 of the target pattern then specifies the mode that the output
must have.
When returning true, set *VECOTYPE_OUT to the vector version of OTYPE.
Also set *VECITYPE_OUT to the vector version of ITYPE if VECITYPE_OUT
is nonnull. */
static bool
vect_supportable_direct_optab_p (vec_info *vinfo, tree otype, tree_code code,
tree itype, tree *vecotype_out,
tree *vecitype_out = NULL)
{
tree vecitype = get_vectype_for_scalar_type (vinfo, itype);
if (!vecitype)
return false;
tree vecotype = get_vectype_for_scalar_type (vinfo, otype);
if (!vecotype)
return false;
optab optab = optab_for_tree_code (code, vecitype, optab_default);
if (!optab)
return false;
insn_code icode = optab_handler (optab, TYPE_MODE (vecitype));
if (icode == CODE_FOR_nothing
|| insn_data[icode].operand[0].mode != TYPE_MODE (vecotype))
return false;
*vecotype_out = vecotype;
if (vecitype_out)
*vecitype_out = vecitype;
return true;
}
/* Round bit precision PRECISION up to a full element. */
static unsigned int
vect_element_precision (unsigned int precision)
{
precision = 1 << ceil_log2 (precision);
return MAX (precision, BITS_PER_UNIT);
}
/* If OP is defined by a statement that's being considered for vectorization,
return information about that statement, otherwise return NULL. */
static stmt_vec_info
vect_get_internal_def (vec_info *vinfo, tree op)
{
stmt_vec_info def_stmt_info = vinfo->lookup_def (op);
if (def_stmt_info
&& STMT_VINFO_DEF_TYPE (def_stmt_info) == vect_internal_def)
return def_stmt_info;
return NULL;
}
/* Check whether NAME, an ssa-name used in STMT_VINFO,
is a result of a type promotion, such that:
DEF_STMT: NAME = NOP (name0)
If CHECK_SIGN is TRUE, check that either both types are signed or both are
unsigned. */
static bool
type_conversion_p (tree name, stmt_vec_info stmt_vinfo, bool check_sign,
tree *orig_type, gimple **def_stmt, bool *promotion)
{
tree type = TREE_TYPE (name);
tree oprnd0;
enum vect_def_type dt;
stmt_vec_info def_stmt_info;
if (!vect_is_simple_use (name, stmt_vinfo->vinfo, &dt, &def_stmt_info,
def_stmt))
return false;
if (dt != vect_internal_def
&& dt != vect_external_def && dt != vect_constant_def)
return false;
if (!*def_stmt)
return false;
if (!is_gimple_assign (*def_stmt))
return false;
if (!CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (*def_stmt)))
return false;
oprnd0 = gimple_assign_rhs1 (*def_stmt);
*orig_type = TREE_TYPE (oprnd0);
if (!INTEGRAL_TYPE_P (type) || !INTEGRAL_TYPE_P (*orig_type)
|| ((TYPE_UNSIGNED (type) != TYPE_UNSIGNED (*orig_type)) && check_sign))
return false;
if (TYPE_PRECISION (type) >= (TYPE_PRECISION (*orig_type) * 2))
*promotion = true;
else
*promotion = false;
if (!vect_is_simple_use (oprnd0, stmt_vinfo->vinfo, &dt))
return false;
return true;
}
/* Holds information about an input operand after some sign changes
and type promotions have been peeled away. */
class vect_unpromoted_value {
public:
vect_unpromoted_value ();
void set_op (tree, vect_def_type, stmt_vec_info = NULL);
/* The value obtained after peeling away zero or more casts. */
tree op;
/* The type of OP. */
tree type;
/* The definition type of OP. */
vect_def_type dt;
/* If OP is the result of peeling at least one cast, and if the cast
of OP itself is a vectorizable statement, CASTER identifies that
statement, otherwise it is null. */
stmt_vec_info caster;
};
inline vect_unpromoted_value::vect_unpromoted_value ()
: op (NULL_TREE),
type (NULL_TREE),
dt (vect_uninitialized_def),
caster (NULL)
{
}
/* Set the operand to OP_IN, its definition type to DT_IN, and the
statement that casts it to CASTER_IN. */
inline void
vect_unpromoted_value::set_op (tree op_in, vect_def_type dt_in,
stmt_vec_info caster_in)
{
op = op_in;
type = TREE_TYPE (op);
dt = dt_in;
caster = caster_in;
}
/* If OP is a vectorizable SSA name, strip a sequence of integer conversions
to reach some vectorizable inner operand OP', continuing as long as it
is possible to convert OP' back to OP using a possible sign change
followed by a possible promotion P. Return this OP', or null if OP is
not a vectorizable SSA name. If there is a promotion P, describe its
input in UNPROM, otherwise describe OP' in UNPROM. If SINGLE_USE_P
is nonnull, set *SINGLE_USE_P to false if any of the SSA names involved
have more than one user.
A successful return means that it is possible to go from OP' to OP
via UNPROM. The cast from OP' to UNPROM is at most a sign change,
whereas the cast from UNPROM to OP might be a promotion, a sign
change, or a nop.
E.g. say we have:
signed short *ptr = ...;
signed short C = *ptr;
unsigned short B = (unsigned short) C; // sign change
signed int A = (signed int) B; // unsigned promotion
...possible other uses of A...
unsigned int OP = (unsigned int) A; // sign change
In this case it's possible to go directly from C to OP using:
OP = (unsigned int) (unsigned short) C;
+------------+ +--------------+
promotion sign change
so OP' would be C. The input to the promotion is B, so UNPROM
would describe B. */
static tree
vect_look_through_possible_promotion (vec_info *vinfo, tree op,
vect_unpromoted_value *unprom,
bool *single_use_p = NULL)
{
tree res = NULL_TREE;
tree op_type = TREE_TYPE (op);
unsigned int orig_precision = TYPE_PRECISION (op_type);
unsigned int min_precision = orig_precision;
stmt_vec_info caster = NULL;
while (TREE_CODE (op) == SSA_NAME && INTEGRAL_TYPE_P (op_type))
{
/* See whether OP is simple enough to vectorize. */
stmt_vec_info def_stmt_info;
gimple *def_stmt;
vect_def_type dt;
if (!vect_is_simple_use (op, vinfo, &dt, &def_stmt_info, &def_stmt))
break;
/* If OP is the input of a demotion, skip over it to see whether
OP is itself the result of a promotion. If so, the combined
effect of the promotion and the demotion might fit the required
pattern, otherwise neither operation fits.
This copes with cases such as the result of an arithmetic
operation being truncated before being stored, and where that
arithmetic operation has been recognized as an over-widened one. */
if (TYPE_PRECISION (op_type) <= min_precision)
{
/* Use OP as the UNPROM described above if we haven't yet
found a promotion, or if using the new input preserves the
sign of the previous promotion. */
if (!res
|| TYPE_PRECISION (unprom->type) == orig_precision
|| TYPE_SIGN (unprom->type) == TYPE_SIGN (op_type))
{
unprom->set_op (op, dt, caster);
min_precision = TYPE_PRECISION (op_type);
}
/* Stop if we've already seen a promotion and if this
conversion does more than change the sign. */
else if (TYPE_PRECISION (op_type)
!= TYPE_PRECISION (unprom->type))
break;
/* The sequence now extends to OP. */
res = op;
}
/* See whether OP is defined by a cast. Record it as CASTER if
the cast is potentially vectorizable. */
if (!def_stmt)
break;
caster = def_stmt_info;
/* Ignore pattern statements, since we don't link uses for them. */
if (caster
&& single_use_p
&& !STMT_VINFO_RELATED_STMT (caster)
&& !has_single_use (res))
*single_use_p = false;
gassign *assign = dyn_cast <gassign *> (def_stmt);
if (!assign || !CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (def_stmt)))
break;
/* Continue with the input to the cast. */
op = gimple_assign_rhs1 (def_stmt);
op_type = TREE_TYPE (op);
}
return res;
}
/* OP is an integer operand to an operation that returns TYPE, and we
want to treat the operation as a widening one. So far we can treat
it as widening from *COMMON_TYPE.
Return true if OP is suitable for such a widening operation,
either widening from *COMMON_TYPE or from some supertype of it.
Update *COMMON_TYPE to the supertype in the latter case.
SHIFT_P is true if OP is a shift amount. */
static bool
vect_joust_widened_integer (tree type, bool shift_p, tree op,
tree *common_type)
{
/* Calculate the minimum precision required by OP, without changing
the sign of either operand. */
unsigned int precision;
if (shift_p)
{
if (!wi::leu_p (wi::to_widest (op), TYPE_PRECISION (type) / 2))
return false;
precision = TREE_INT_CST_LOW (op);
}
else
{
precision = wi::min_precision (wi::to_widest (op),
TYPE_SIGN (*common_type));
if (precision * 2 > TYPE_PRECISION (type))
return false;
}
/* If OP requires a wider type, switch to that type. The checks
above ensure that this is still narrower than the result. */
precision = vect_element_precision (precision);
if (TYPE_PRECISION (*common_type) < precision)
*common_type = build_nonstandard_integer_type
(precision, TYPE_UNSIGNED (*common_type));
return true;
}
/* Return true if the common supertype of NEW_TYPE and *COMMON_TYPE
is narrower than type, storing the supertype in *COMMON_TYPE if so. */
static bool
vect_joust_widened_type (tree type, tree new_type, tree *common_type)
{
if (types_compatible_p (*common_type, new_type))
return true;
/* See if *COMMON_TYPE can hold all values of NEW_TYPE. */
if ((TYPE_PRECISION (new_type) < TYPE_PRECISION (*common_type))
&& (TYPE_UNSIGNED (new_type) || !TYPE_UNSIGNED (*common_type)))
return true;
/* See if NEW_TYPE can hold all values of *COMMON_TYPE. */
if (TYPE_PRECISION (*common_type) < TYPE_PRECISION (new_type)
&& (TYPE_UNSIGNED (*common_type) || !TYPE_UNSIGNED (new_type)))
{
*common_type = new_type;
return true;
}
/* We have mismatched signs, with the signed type being
no wider than the unsigned type. In this case we need
a wider signed type. */
unsigned int precision = MAX (TYPE_PRECISION (*common_type),
TYPE_PRECISION (new_type));
precision *= 2;
if (precision * 2 > TYPE_PRECISION (type))
return false;
*common_type = build_nonstandard_integer_type (precision, false);
return true;
}
/* Check whether STMT_INFO can be viewed as a tree of integer operations
in which each node either performs CODE or WIDENED_CODE, and where
each leaf operand is narrower than the result of STMT_INFO. MAX_NOPS
specifies the maximum number of leaf operands. SHIFT_P says whether
CODE and WIDENED_CODE are some sort of shift.
If STMT_INFO is such a tree, return the number of leaf operands
and describe them in UNPROM[0] onwards. Also set *COMMON_TYPE
to a type that (a) is narrower than the result of STMT_INFO and
(b) can hold all leaf operand values.
Return 0 if STMT_INFO isn't such a tree, or if no such COMMON_TYPE
exists. */
static unsigned int
vect_widened_op_tree (stmt_vec_info stmt_info, tree_code code,
tree_code widened_code, bool shift_p,
unsigned int max_nops,
vect_unpromoted_value *unprom, tree *common_type)
{
/* Check for an integer operation with the right code. */
vec_info *vinfo = stmt_info->vinfo;
gassign *assign = dyn_cast <gassign *> (stmt_info->stmt);
if (!assign)
return 0;
tree_code rhs_code = gimple_assign_rhs_code (assign);
if (rhs_code != code && rhs_code != widened_code)
return 0;
tree type = gimple_expr_type (assign);
if (!INTEGRAL_TYPE_P (type))
return 0;
/* Assume that both operands will be leaf operands. */
max_nops -= 2;
/* Check the operands. */
unsigned int next_op = 0;
for (unsigned int i = 0; i < 2; ++i)
{
vect_unpromoted_value *this_unprom = &unprom[next_op];
unsigned int nops = 1;
tree op = gimple_op (assign, i + 1);
if (i == 1 && TREE_CODE (op) == INTEGER_CST)
{
/* We already have a common type from earlier operands.
Update it to account for OP. */
this_unprom->set_op (op, vect_constant_def);
if (!vect_joust_widened_integer (type, shift_p, op, common_type))
return 0;
}
else
{
/* Only allow shifts by constants. */
if (shift_p && i == 1)
return 0;
if (!vect_look_through_possible_promotion (stmt_info->vinfo, op,
this_unprom))
return 0;
if (TYPE_PRECISION (this_unprom->type) == TYPE_PRECISION (type))
{
/* The operand isn't widened. If STMT_INFO has the code
for an unwidened operation, recursively check whether
this operand is a node of the tree. */
if (rhs_code != code
|| max_nops == 0
|| this_unprom->dt != vect_internal_def)
return 0;
/* Give back the leaf slot allocated above now that we're
not treating this as a leaf operand. */
max_nops += 1;
/* Recursively process the definition of the operand. */
stmt_vec_info def_stmt_info
= vinfo->lookup_def (this_unprom->op);
nops = vect_widened_op_tree (def_stmt_info, code, widened_code,
shift_p, max_nops, this_unprom,
common_type);
if (nops == 0)
return 0;
max_nops -= nops;
}
else
{
/* Make sure that the operand is narrower than the result. */
if (TYPE_PRECISION (this_unprom->type) * 2
> TYPE_PRECISION (type))
return 0;
/* Update COMMON_TYPE for the new operand. */
if (i == 0)
*common_type = this_unprom->type;
else if (!vect_joust_widened_type (type, this_unprom->type,
common_type))
return 0;
}
}
next_op += nops;
}
return next_op;
}
/* Helper to return a new temporary for pattern of TYPE for STMT. If STMT
is NULL, the caller must set SSA_NAME_DEF_STMT for the returned SSA var. */
static tree
vect_recog_temp_ssa_var (tree type, gimple *stmt)
{
return make_temp_ssa_name (type, stmt, "patt");
}
/* STMT2_INFO describes a type conversion that could be split into STMT1
followed by a version of STMT2_INFO that takes NEW_RHS as its first
input. Try to do this using pattern statements, returning true on
success. */
static bool
vect_split_statement (stmt_vec_info stmt2_info, tree new_rhs,
gimple *stmt1, tree vectype)
{
vec_info *vinfo = stmt2_info->vinfo;
if (is_pattern_stmt_p (stmt2_info))
{
/* STMT2_INFO is part of a pattern. Get the statement to which
the pattern is attached. */
stmt_vec_info orig_stmt2_info = STMT_VINFO_RELATED_STMT (stmt2_info);
vect_init_pattern_stmt (stmt1, orig_stmt2_info, vectype);
if (dump_enabled_p ())
dump_printf_loc (MSG_NOTE, vect_location,
"Splitting pattern statement: %G", stmt2_info->stmt);
/* Since STMT2_INFO is a pattern statement, we can change it
in-situ without worrying about changing the code for the
containing block. */
gimple_assign_set_rhs1 (stmt2_info->stmt, new_rhs);
if (dump_enabled_p ())
{
dump_printf_loc (MSG_NOTE, vect_location, "into: %G", stmt1);
dump_printf_loc (MSG_NOTE, vect_location, "and: %G",
stmt2_info->stmt);
}
gimple_seq *def_seq = &STMT_VINFO_PATTERN_DEF_SEQ (orig_stmt2_info);
if (STMT_VINFO_RELATED_STMT (orig_stmt2_info) == stmt2_info)
/* STMT2_INFO is the actual pattern statement. Add STMT1
to the end of the definition sequence. */
gimple_seq_add_stmt_without_update (def_seq, stmt1);
else
{
/* STMT2_INFO belongs to the definition sequence. Insert STMT1
before it. */
gimple_stmt_iterator gsi = gsi_for_stmt (stmt2_info->stmt, def_seq);
gsi_insert_before_without_update (&gsi, stmt1, GSI_SAME_STMT);
}
return true;
}
else
{
/* STMT2_INFO doesn't yet have a pattern. Try to create a
two-statement pattern now. */
gcc_assert (!STMT_VINFO_RELATED_STMT (stmt2_info));
tree lhs_type = TREE_TYPE (gimple_get_lhs (stmt2_info->stmt));
tree lhs_vectype = get_vectype_for_scalar_type (vinfo, lhs_type);
if (!lhs_vectype)
return false;
if (dump_enabled_p ())
dump_printf_loc (MSG_NOTE, vect_location,
"Splitting statement: %G", stmt2_info->stmt);
/* Add STMT1 as a singleton pattern definition sequence. */
gimple_seq *def_seq = &STMT_VINFO_PATTERN_DEF_SEQ (stmt2_info);
vect_init_pattern_stmt (stmt1, stmt2_info, vectype);
gimple_seq_add_stmt_without_update (def_seq, stmt1);
/* Build the second of the two pattern statements. */
tree new_lhs = vect_recog_temp_ssa_var (lhs_type, NULL);
gassign *new_stmt2 = gimple_build_assign (new_lhs, NOP_EXPR, new_rhs);
vect_set_pattern_stmt (new_stmt2, stmt2_info, lhs_vectype);
if (dump_enabled_p ())
{
dump_printf_loc (MSG_NOTE, vect_location,
"into pattern statements: %G", stmt1);
dump_printf_loc (MSG_NOTE, vect_location, "and: %G", new_stmt2);
}
return true;
}
}
/* Convert UNPROM to TYPE and return the result, adding new statements
to STMT_INFO's pattern definition statements if no better way is
available. VECTYPE is the vector form of TYPE. */
static tree
vect_convert_input (stmt_vec_info stmt_info, tree type,
vect_unpromoted_value *unprom, tree vectype)
{
vec_info *vinfo = stmt_info->vinfo;
/* Check for a no-op conversion. */
if (types_compatible_p (type, TREE_TYPE (unprom->op)))
return unprom->op;
/* Allow the caller to create constant vect_unpromoted_values. */
if (TREE_CODE (unprom->op) == INTEGER_CST)
return wide_int_to_tree (type, wi::to_widest (unprom->op));
tree input = unprom->op;
if (unprom->caster)
{
tree lhs = gimple_get_lhs (unprom->caster->stmt);
tree lhs_type = TREE_TYPE (lhs);
/* If the result of the existing cast is the right width, use it
instead of the source of the cast. */
if (TYPE_PRECISION (lhs_type) == TYPE_PRECISION (type))
input = lhs;
/* If the precision we want is between the source and result
precisions of the existing cast, try splitting the cast into
two and tapping into a mid-way point. */
else if (TYPE_PRECISION (lhs_type) > TYPE_PRECISION (type)
&& TYPE_PRECISION (type) > TYPE_PRECISION (unprom->type))
{
/* In order to preserve the semantics of the original cast,
give the mid-way point the same signedness as the input value.
It would be possible to use a signed type here instead if
TYPE is signed and UNPROM->TYPE is unsigned, but that would
make the sign of the midtype sensitive to the order in
which we process the statements, since the signedness of
TYPE is the signedness required by just one of possibly
many users. Also, unsigned promotions are usually as cheap
as or cheaper than signed ones, so it's better to keep an
unsigned promotion. */
tree midtype = build_nonstandard_integer_type
(TYPE_PRECISION (type), TYPE_UNSIGNED (unprom->type));
tree vec_midtype = get_vectype_for_scalar_type (vinfo, midtype);
if (vec_midtype)
{
input = vect_recog_temp_ssa_var (midtype, NULL);
gassign *new_stmt = gimple_build_assign (input, NOP_EXPR,
unprom->op);
if (!vect_split_statement (unprom->caster, input, new_stmt,
vec_midtype))
append_pattern_def_seq (stmt_info, new_stmt, vec_midtype);
}
}
/* See if we can reuse an existing result. */
if (types_compatible_p (type, TREE_TYPE (input)))
return input;
}
/* We need a new conversion statement. */
tree new_op = vect_recog_temp_ssa_var (type, NULL);
gassign *new_stmt = gimple_build_assign (new_op, NOP_EXPR, input);
/* If OP is an external value, see if we can insert the new statement
on an incoming edge. */
if (input == unprom->op && unprom->dt == vect_external_def)
if (edge e = vect_get_external_def_edge (stmt_info->vinfo, input))
{
basic_block new_bb = gsi_insert_on_edge_immediate (e, new_stmt);
gcc_assert (!new_bb);
return new_op;
}
/* As a (common) last resort, add the statement to the pattern itself. */
append_pattern_def_seq (stmt_info, new_stmt, vectype);
return new_op;
}
/* Invoke vect_convert_input for N elements of UNPROM and store the
result in the corresponding elements of RESULT. */
static void
vect_convert_inputs (stmt_vec_info stmt_info, unsigned int n,
tree *result, tree type, vect_unpromoted_value *unprom,
tree vectype)
{
for (unsigned int i = 0; i < n; ++i)
{
unsigned int j;
for (j = 0; j < i; ++j)
if (unprom[j].op == unprom[i].op)
break;
if (j < i)
result[i] = result[j];
else
result[i] = vect_convert_input (stmt_info, type, &unprom[i], vectype);
}
}
/* The caller has created a (possibly empty) sequence of pattern definition
statements followed by a single statement PATTERN_STMT. Cast the result
of this final statement to TYPE. If a new statement is needed, add
PATTERN_STMT to the end of STMT_INFO's pattern definition statements
and return the new statement, otherwise return PATTERN_STMT as-is.
VECITYPE is the vector form of PATTERN_STMT's result type. */
static gimple *
vect_convert_output (stmt_vec_info stmt_info, tree type, gimple *pattern_stmt,
tree vecitype)
{
tree lhs = gimple_get_lhs (pattern_stmt);
if (!types_compatible_p (type, TREE_TYPE (lhs)))
{
append_pattern_def_seq (stmt_info, pattern_stmt, vecitype);
tree cast_var = vect_recog_temp_ssa_var (type, NULL);
pattern_stmt = gimple_build_assign (cast_var, NOP_EXPR, lhs);
}
return pattern_stmt;
}
/* Return true if STMT_VINFO describes a reduction for which reassociation
is allowed. If STMT_INFO is part of a group, assume that it's part of
a reduction chain and optimistically assume that all statements
except the last allow reassociation.
Also require it to have code CODE and to be a reduction
in the outermost loop. When returning true, store the operands in
*OP0_OUT and *OP1_OUT. */
static bool
vect_reassociating_reduction_p (stmt_vec_info stmt_info, tree_code code,
tree *op0_out, tree *op1_out)
{
loop_vec_info loop_info = STMT_VINFO_LOOP_VINFO (stmt_info);
if (!loop_info)
return false;
gassign *assign = dyn_cast <gassign *> (stmt_info->stmt);
if (!assign || gimple_assign_rhs_code (assign) != code)
return false;
/* We don't allow changing the order of the computation in the inner-loop
when doing outer-loop vectorization. */
class loop *loop = LOOP_VINFO_LOOP (loop_info);
if (loop && nested_in_vect_loop_p (loop, stmt_info))
return false;
if (STMT_VINFO_DEF_TYPE (stmt_info) == vect_reduction_def)
{
if (needs_fold_left_reduction_p (TREE_TYPE (gimple_assign_lhs (assign)),
code))
return false;
}
else if (REDUC_GROUP_FIRST_ELEMENT (stmt_info) == NULL)
return false;
*op0_out = gimple_assign_rhs1 (assign);
*op1_out = gimple_assign_rhs2 (assign);
if (commutative_tree_code (code) && STMT_VINFO_REDUC_IDX (stmt_info) == 0)
std::swap (*op0_out, *op1_out);
return true;
}
/* Function vect_recog_dot_prod_pattern
Try to find the following pattern:
type x_t, y_t;
TYPE1 prod;
TYPE2 sum = init;
loop:
sum_0 = phi <init, sum_1>
S1 x_t = ...
S2 y_t = ...
S3 x_T = (TYPE1) x_t;
S4 y_T = (TYPE1) y_t;
S5 prod = x_T * y_T;
[S6 prod = (TYPE2) prod; #optional]
S7 sum_1 = prod + sum_0;
where 'TYPE1' is exactly double the size of type 'type', and 'TYPE2' is the
same size of 'TYPE1' or bigger. This is a special case of a reduction
computation.
Input:
* STMT_VINFO: The stmt from which the pattern search begins. In the
example, when this function is called with S7, the pattern {S3,S4,S5,S6,S7}
will be detected.
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace the sequence of
stmts that constitute the pattern. In this case it will be:
WIDEN_DOT_PRODUCT <x_t, y_t, sum_0>
Note: The dot-prod idiom is a widening reduction pattern that is
vectorized without preserving all the intermediate results. It
produces only N/2 (widened) results (by summing up pairs of
intermediate results) rather than all N results. Therefore, we
cannot allow this pattern when we want to get all the results and in
the correct order (as is the case when this computation is in an
inner-loop nested in an outer-loop that us being vectorized). */
static gimple *
vect_recog_dot_prod_pattern (stmt_vec_info stmt_vinfo, tree *type_out)
{
tree oprnd0, oprnd1;
gimple *last_stmt = stmt_vinfo->stmt;
vec_info *vinfo = stmt_vinfo->vinfo;
tree type, half_type;
gimple *pattern_stmt;
tree var;
/* Look for the following pattern
DX = (TYPE1) X;
DY = (TYPE1) Y;
DPROD = DX * DY;
DDPROD = (TYPE2) DPROD;
sum_1 = DDPROD + sum_0;
In which
- DX is double the size of X
- DY is double the size of Y
- DX, DY, DPROD all have the same type
- sum is the same size of DPROD or bigger
- sum has been recognized as a reduction variable.
This is equivalent to:
DPROD = X w* Y; #widen mult
sum_1 = DPROD w+ sum_0; #widen summation
or
DPROD = X w* Y; #widen mult
sum_1 = DPROD + sum_0; #summation
*/
/* Starting from LAST_STMT, follow the defs of its uses in search
of the above pattern. */
if (!vect_reassociating_reduction_p (stmt_vinfo, PLUS_EXPR,
&oprnd0, &oprnd1))
return NULL;
type = gimple_expr_type (last_stmt);
vect_unpromoted_value unprom_mult;
oprnd0 = vect_look_through_possible_promotion (vinfo, oprnd0, &unprom_mult);
/* So far so good. Since last_stmt was detected as a (summation) reduction,
we know that oprnd1 is the reduction variable (defined by a loop-header
phi), and oprnd0 is an ssa-name defined by a stmt in the loop body.
Left to check that oprnd0 is defined by a (widen_)mult_expr */
if (!oprnd0)
return NULL;
stmt_vec_info mult_vinfo = vect_get_internal_def (vinfo, oprnd0);
if (!mult_vinfo)
return NULL;
/* FORNOW. Can continue analyzing the def-use chain when this stmt in a phi
inside the loop (in case we are analyzing an outer-loop). */
vect_unpromoted_value unprom0[2];
if (!vect_widened_op_tree (mult_vinfo, MULT_EXPR, WIDEN_MULT_EXPR,
false, 2, unprom0, &half_type))
return NULL;
/* If there are two widening operations, make sure they agree on
the sign of the extension. */
if (TYPE_PRECISION (unprom_mult.type) != TYPE_PRECISION (type)
&& TYPE_SIGN (unprom_mult.type) != TYPE_SIGN (half_type))
return NULL;
vect_pattern_detected ("vect_recog_dot_prod_pattern", last_stmt);
tree half_vectype;
if (!vect_supportable_direct_optab_p (vinfo, type, DOT_PROD_EXPR, half_type,
type_out, &half_vectype))
return NULL;
/* Get the inputs in the appropriate types. */
tree mult_oprnd[2];
vect_convert_inputs (stmt_vinfo, 2, mult_oprnd, half_type,
unprom0, half_vectype);
var = vect_recog_temp_ssa_var (type, NULL);
pattern_stmt = gimple_build_assign (var, DOT_PROD_EXPR,
mult_oprnd[0], mult_oprnd[1], oprnd1);
return pattern_stmt;
}
/* Function vect_recog_sad_pattern
Try to find the following Sum of Absolute Difference (SAD) pattern:
type x_t, y_t;
signed TYPE1 diff, abs_diff;
TYPE2 sum = init;
loop:
sum_0 = phi <init, sum_1>
S1 x_t = ...
S2 y_t = ...
S3 x_T = (TYPE1) x_t;
S4 y_T = (TYPE1) y_t;
S5 diff = x_T - y_T;
S6 abs_diff = ABS_EXPR <diff>;
[S7 abs_diff = (TYPE2) abs_diff; #optional]
S8 sum_1 = abs_diff + sum_0;
where 'TYPE1' is at least double the size of type 'type', and 'TYPE2' is the
same size of 'TYPE1' or bigger. This is a special case of a reduction
computation.
Input:
* STMT_VINFO: The stmt from which the pattern search begins. In the
example, when this function is called with S8, the pattern
{S3,S4,S5,S6,S7,S8} will be detected.
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace the sequence of
stmts that constitute the pattern. In this case it will be:
SAD_EXPR <x_t, y_t, sum_0>
*/
static gimple *
vect_recog_sad_pattern (stmt_vec_info stmt_vinfo, tree *type_out)
{
gimple *last_stmt = stmt_vinfo->stmt;
vec_info *vinfo = stmt_vinfo->vinfo;
tree half_type;
/* Look for the following pattern
DX = (TYPE1) X;
DY = (TYPE1) Y;
DDIFF = DX - DY;
DAD = ABS_EXPR <DDIFF>;
DDPROD = (TYPE2) DPROD;
sum_1 = DAD + sum_0;
In which
- DX is at least double the size of X
- DY is at least double the size of Y
- DX, DY, DDIFF, DAD all have the same type
- sum is the same size of DAD or bigger
- sum has been recognized as a reduction variable.
This is equivalent to:
DDIFF = X w- Y; #widen sub
DAD = ABS_EXPR <DDIFF>;
sum_1 = DAD w+ sum_0; #widen summation
or
DDIFF = X w- Y; #widen sub
DAD = ABS_EXPR <DDIFF>;
sum_1 = DAD + sum_0; #summation
*/
/* Starting from LAST_STMT, follow the defs of its uses in search
of the above pattern. */
tree plus_oprnd0, plus_oprnd1;
if (!vect_reassociating_reduction_p (stmt_vinfo, PLUS_EXPR,
&plus_oprnd0, &plus_oprnd1))
return NULL;
tree sum_type = gimple_expr_type (last_stmt);
/* Any non-truncating sequence of conversions is OK here, since
with a successful match, the result of the ABS(U) is known to fit
within the nonnegative range of the result type. (It cannot be the
negative of the minimum signed value due to the range of the widening
MINUS_EXPR.) */
vect_unpromoted_value unprom_abs;
plus_oprnd0 = vect_look_through_possible_promotion (vinfo, plus_oprnd0,
&unprom_abs);
/* So far so good. Since last_stmt was detected as a (summation) reduction,
we know that plus_oprnd1 is the reduction variable (defined by a loop-header
phi), and plus_oprnd0 is an ssa-name defined by a stmt in the loop body.
Then check that plus_oprnd0 is defined by an abs_expr. */
if (!plus_oprnd0)
return NULL;
stmt_vec_info abs_stmt_vinfo = vect_get_internal_def (vinfo, plus_oprnd0);
if (!abs_stmt_vinfo)
return NULL;
/* FORNOW. Can continue analyzing the def-use chain when this stmt in a phi
inside the loop (in case we are analyzing an outer-loop). */
gassign *abs_stmt = dyn_cast <gassign *> (abs_stmt_vinfo->stmt);
if (!abs_stmt
|| (gimple_assign_rhs_code (abs_stmt) != ABS_EXPR
&& gimple_assign_rhs_code (abs_stmt) != ABSU_EXPR))
return NULL;
tree abs_oprnd = gimple_assign_rhs1 (abs_stmt);
tree abs_type = TREE_TYPE (abs_oprnd);
if (TYPE_UNSIGNED (abs_type))
return NULL;
/* Peel off conversions from the ABS input. This can involve sign
changes (e.g. from an unsigned subtraction to a signed ABS input)
or signed promotion, but it can't include unsigned promotion.
(Note that ABS of an unsigned promotion should have been folded
away before now anyway.) */
vect_unpromoted_value unprom_diff;
abs_oprnd = vect_look_through_possible_promotion (vinfo, abs_oprnd,
&unprom_diff);
if (!abs_oprnd)
return NULL;
if (TYPE_PRECISION (unprom_diff.type) != TYPE_PRECISION (abs_type)
&& TYPE_UNSIGNED (unprom_diff.type))
return NULL;
/* We then detect if the operand of abs_expr is defined by a minus_expr. */
stmt_vec_info diff_stmt_vinfo = vect_get_internal_def (vinfo, abs_oprnd);
if (!diff_stmt_vinfo)
return NULL;
/* FORNOW. Can continue analyzing the def-use chain when this stmt in a phi
inside the loop (in case we are analyzing an outer-loop). */
vect_unpromoted_value unprom[2];
if (!vect_widened_op_tree (diff_stmt_vinfo, MINUS_EXPR, MINUS_EXPR,
false, 2, unprom, &half_type))
return NULL;
vect_pattern_detected ("vect_recog_sad_pattern", last_stmt);
tree half_vectype;
if (!vect_supportable_direct_optab_p (vinfo, sum_type, SAD_EXPR, half_type,
type_out, &half_vectype))
return NULL;
/* Get the inputs to the SAD_EXPR in the appropriate types. */
tree sad_oprnd[2];
vect_convert_inputs (stmt_vinfo, 2, sad_oprnd, half_type,
unprom, half_vectype);
tree var = vect_recog_temp_ssa_var (sum_type, NULL);
gimple *pattern_stmt = gimple_build_assign (var, SAD_EXPR, sad_oprnd[0],
sad_oprnd[1], plus_oprnd1);
return pattern_stmt;
}
/* Recognize an operation that performs ORIG_CODE on widened inputs,
so that it can be treated as though it had the form:
A_TYPE a;
B_TYPE b;
HALF_TYPE a_cast = (HALF_TYPE) a; // possible no-op
HALF_TYPE b_cast = (HALF_TYPE) b; // possible no-op
| RES_TYPE a_extend = (RES_TYPE) a_cast; // promotion from HALF_TYPE
| RES_TYPE b_extend = (RES_TYPE) b_cast; // promotion from HALF_TYPE
| RES_TYPE res = a_extend ORIG_CODE b_extend;
Try to replace the pattern with:
A_TYPE a;
B_TYPE b;
HALF_TYPE a_cast = (HALF_TYPE) a; // possible no-op
HALF_TYPE b_cast = (HALF_TYPE) b; // possible no-op
| EXT_TYPE ext = a_cast WIDE_CODE b_cast;
| RES_TYPE res = (EXT_TYPE) ext; // possible no-op
where EXT_TYPE is wider than HALF_TYPE but has the same signedness.
SHIFT_P is true if ORIG_CODE and WIDE_CODE are shifts. NAME is the
name of the pattern being matched, for dump purposes. */
static gimple *
vect_recog_widen_op_pattern (stmt_vec_info last_stmt_info, tree *type_out,
tree_code orig_code, tree_code wide_code,
bool shift_p, const char *name)
{
vec_info *vinfo = last_stmt_info->vinfo;
gimple *last_stmt = last_stmt_info->stmt;
vect_unpromoted_value unprom[2];
tree half_type;
if (!vect_widened_op_tree (last_stmt_info, orig_code, orig_code,
shift_p, 2, unprom, &half_type))
return NULL;
/* Pattern detected. */
vect_pattern_detected (name, last_stmt);
tree type = gimple_expr_type (last_stmt);
tree itype = type;
if (TYPE_PRECISION (type) != TYPE_PRECISION (half_type) * 2
|| TYPE_UNSIGNED (type) != TYPE_UNSIGNED (half_type))
itype = build_nonstandard_integer_type (TYPE_PRECISION (half_type) * 2,
TYPE_UNSIGNED (half_type));
/* Check target support */
tree vectype = get_vectype_for_scalar_type (vinfo, half_type);
tree vecitype = get_vectype_for_scalar_type (vinfo, itype);
enum tree_code dummy_code;
int dummy_int;
auto_vec<tree> dummy_vec;
if (!vectype
|| !vecitype
|| !supportable_widening_operation (wide_code, last_stmt_info,
vecitype, vectype,
&dummy_code, &dummy_code,
&dummy_int, &dummy_vec))
return NULL;
*type_out = get_vectype_for_scalar_type (vinfo, type);
if (!*type_out)
return NULL;
tree oprnd[2];
vect_convert_inputs (last_stmt_info, 2, oprnd, half_type, unprom, vectype);
tree var = vect_recog_temp_ssa_var (itype, NULL);
gimple *pattern_stmt = gimple_build_assign (var, wide_code,
oprnd[0], oprnd[1]);
return vect_convert_output (last_stmt_info, type, pattern_stmt, vecitype);
}
/* Try to detect multiplication on widened inputs, converting MULT_EXPR
to WIDEN_MULT_EXPR. See vect_recog_widen_op_pattern for details. */
static gimple *
vect_recog_widen_mult_pattern (stmt_vec_info last_stmt_info, tree *type_out)
{
return vect_recog_widen_op_pattern (last_stmt_info, type_out, MULT_EXPR,
WIDEN_MULT_EXPR, false,
"vect_recog_widen_mult_pattern");
}
/* Function vect_recog_pow_pattern
Try to find the following pattern:
x = POW (y, N);
with POW being one of pow, powf, powi, powif and N being
either 2 or 0.5.
Input:
* STMT_VINFO: The stmt from which the pattern search begins.
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace the sequence of
stmts that constitute the pattern. In this case it will be:
x = x * x
or
x = sqrt (x)
*/
static gimple *
vect_recog_pow_pattern (stmt_vec_info stmt_vinfo, tree *type_out)
{
vec_info *vinfo = stmt_vinfo->vinfo;
gimple *last_stmt = stmt_vinfo->stmt;
tree base, exp;
gimple *stmt;
tree var;
if (!is_gimple_call (last_stmt) || gimple_call_lhs (last_stmt) == NULL)
return NULL;
switch (gimple_call_combined_fn (last_stmt))
{
CASE_CFN_POW:
CASE_CFN_POWI:
break;
default:
return NULL;
}
base = gimple_call_arg (last_stmt, 0);
exp = gimple_call_arg (last_stmt, 1);
if (TREE_CODE (exp) != REAL_CST
&& TREE_CODE (exp) != INTEGER_CST)
{
if (flag_unsafe_math_optimizations
&& TREE_CODE (base) == REAL_CST
&& gimple_call_builtin_p (last_stmt, BUILT_IN_NORMAL))
{
combined_fn log_cfn;
built_in_function exp_bfn;
switch (DECL_FUNCTION_CODE (gimple_call_fndecl (last_stmt)))
{
case BUILT_IN_POW:
log_cfn = CFN_BUILT_IN_LOG;
exp_bfn = BUILT_IN_EXP;
break;
case BUILT_IN_POWF:
log_cfn = CFN_BUILT_IN_LOGF;
exp_bfn = BUILT_IN_EXPF;
break;
case BUILT_IN_POWL:
log_cfn = CFN_BUILT_IN_LOGL;
exp_bfn = BUILT_IN_EXPL;
break;
default:
return NULL;
}
tree logc = fold_const_call (log_cfn, TREE_TYPE (base), base);
tree exp_decl = builtin_decl_implicit (exp_bfn);
/* Optimize pow (C, x) as exp (log (C) * x). Normally match.pd
does that, but if C is a power of 2, we want to use
exp2 (log2 (C) * x) in the non-vectorized version, but for
vectorization we don't have vectorized exp2. */
if (logc
&& TREE_CODE (logc) == REAL_CST
&& exp_decl
&& lookup_attribute ("omp declare simd",
DECL_ATTRIBUTES (exp_decl)))
{
cgraph_node *node = cgraph_node::get_create (exp_decl);
if (node->simd_clones == NULL)
{
if (targetm.simd_clone.compute_vecsize_and_simdlen == NULL
|| node->definition)
return NULL;
expand_simd_clones (node);
if (node->simd_clones == NULL)
return NULL;
}
*type_out = get_vectype_for_scalar_type (vinfo, TREE_TYPE (base));
if (!*type_out)
return NULL;
tree def = vect_recog_temp_ssa_var (TREE_TYPE (base), NULL);
gimple *g = gimple_build_assign (def, MULT_EXPR, exp, logc);
append_pattern_def_seq (stmt_vinfo, g);
tree res = vect_recog_temp_ssa_var (TREE_TYPE (base), NULL);
g = gimple_build_call (exp_decl, 1, def);
gimple_call_set_lhs (g, res);
return g;
}
}
return NULL;
}
/* We now have a pow or powi builtin function call with a constant
exponent. */
/* Catch squaring. */
if ((tree_fits_shwi_p (exp)
&& tree_to_shwi (exp) == 2)
|| (TREE_CODE (exp) == REAL_CST
&& real_equal (&TREE_REAL_CST (exp), &dconst2)))
{
if (!vect_supportable_direct_optab_p (vinfo, TREE_TYPE (base), MULT_EXPR,
TREE_TYPE (base), type_out))
return NULL;
var = vect_recog_temp_ssa_var (TREE_TYPE (base), NULL);
stmt = gimple_build_assign (var, MULT_EXPR, base, base);
return stmt;
}
/* Catch square root. */
if (TREE_CODE (exp) == REAL_CST
&& real_equal (&TREE_REAL_CST (exp), &dconsthalf))
{
*type_out = get_vectype_for_scalar_type (vinfo, TREE_TYPE (base));
if (*type_out
&& direct_internal_fn_supported_p (IFN_SQRT, *type_out,
OPTIMIZE_FOR_SPEED))
{
gcall *stmt = gimple_build_call_internal (IFN_SQRT, 1, base);
var = vect_recog_temp_ssa_var (TREE_TYPE (base), stmt);
gimple_call_set_lhs (stmt, var);
gimple_call_set_nothrow (stmt, true);
return stmt;
}
}
return NULL;
}
/* Function vect_recog_widen_sum_pattern
Try to find the following pattern:
type x_t;
TYPE x_T, sum = init;
loop:
sum_0 = phi <init, sum_1>
S1 x_t = *p;
S2 x_T = (TYPE) x_t;
S3 sum_1 = x_T + sum_0;
where type 'TYPE' is at least double the size of type 'type', i.e - we're
summing elements of type 'type' into an accumulator of type 'TYPE'. This is
a special case of a reduction computation.
Input:
* STMT_VINFO: The stmt from which the pattern search begins. In the example,
when this function is called with S3, the pattern {S2,S3} will be detected.
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace the sequence of
stmts that constitute the pattern. In this case it will be:
WIDEN_SUM <x_t, sum_0>
Note: The widening-sum idiom is a widening reduction pattern that is
vectorized without preserving all the intermediate results. It
produces only N/2 (widened) results (by summing up pairs of
intermediate results) rather than all N results. Therefore, we
cannot allow this pattern when we want to get all the results and in
the correct order (as is the case when this computation is in an
inner-loop nested in an outer-loop that us being vectorized). */
static gimple *
vect_recog_widen_sum_pattern (stmt_vec_info stmt_vinfo, tree *type_out)
{
gimple *last_stmt = stmt_vinfo->stmt;
tree oprnd0, oprnd1;
vec_info *vinfo = stmt_vinfo->vinfo;
tree type;
gimple *pattern_stmt;
tree var;
/* Look for the following pattern
DX = (TYPE) X;
sum_1 = DX + sum_0;
In which DX is at least double the size of X, and sum_1 has been
recognized as a reduction variable.
*/
/* Starting from LAST_STMT, follow the defs of its uses in search
of the above pattern. */
if (!vect_reassociating_reduction_p (stmt_vinfo, PLUS_EXPR,
&oprnd0, &oprnd1))
return NULL;
type = gimple_expr_type (last_stmt);
/* So far so good. Since last_stmt was detected as a (summation) reduction,
we know that oprnd1 is the reduction variable (defined by a loop-header
phi), and oprnd0 is an ssa-name defined by a stmt in the loop body.
Left to check that oprnd0 is defined by a cast from type 'type' to type
'TYPE'. */
vect_unpromoted_value unprom0;
if (!vect_look_through_possible_promotion (vinfo, oprnd0, &unprom0)
|| TYPE_PRECISION (unprom0.type) * 2 > TYPE_PRECISION (type))
return NULL;
vect_pattern_detected ("vect_recog_widen_sum_pattern", last_stmt);
if (!vect_supportable_direct_optab_p (vinfo, type, WIDEN_SUM_EXPR,
unprom0.type, type_out))
return NULL;
var = vect_recog_temp_ssa_var (type, NULL);
pattern_stmt = gimple_build_assign (var, WIDEN_SUM_EXPR, unprom0.op, oprnd1);
return pattern_stmt;
}
/* Recognize cases in which an operation is performed in one type WTYPE
but could be done more efficiently in a narrower type NTYPE. For example,
if we have:
ATYPE a; // narrower than NTYPE
BTYPE b; // narrower than NTYPE
WTYPE aw = (WTYPE) a;
WTYPE bw = (WTYPE) b;
WTYPE res = aw + bw; // only uses of aw and bw
then it would be more efficient to do:
NTYPE an = (NTYPE) a;
NTYPE bn = (NTYPE) b;
NTYPE resn = an + bn;
WTYPE res = (WTYPE) resn;
Other situations include things like:
ATYPE a; // NTYPE or narrower
WTYPE aw = (WTYPE) a;
WTYPE res = aw + b;
when only "(NTYPE) res" is significant. In that case it's more efficient
to truncate "b" and do the operation on NTYPE instead:
NTYPE an = (NTYPE) a;
NTYPE bn = (NTYPE) b; // truncation
NTYPE resn = an + bn;
WTYPE res = (WTYPE) resn;
All users of "res" should then use "resn" instead, making the final
statement dead (not marked as relevant). The final statement is still
needed to maintain the type correctness of the IR.
vect_determine_precisions has already determined the minimum
precison of the operation and the minimum precision required
by users of the result. */
static gimple *
vect_recog_over_widening_pattern (stmt_vec_info last_stmt_info, tree *type_out)
{
gassign *last_stmt = dyn_cast <gassign *> (last_stmt_info->stmt);
if (!last_stmt)
return NULL;
/* See whether we have found that this operation can be done on a
narrower type without changing its semantics. */
unsigned int new_precision = last_stmt_info->operation_precision;
if (!new_precision)
return NULL;
vec_info *vinfo = last_stmt_info->vinfo;
tree lhs = gimple_assign_lhs (last_stmt);
tree type = TREE_TYPE (lhs);
tree_code code = gimple_assign_rhs_code (last_stmt);
/* Keep the first operand of a COND_EXPR as-is: only the other two
operands are interesting. */
unsigned int first_op = (code == COND_EXPR ? 2 : 1);
/* Check the operands. */
unsigned int nops = gimple_num_ops (last_stmt) - first_op;
auto_vec <vect_unpromoted_value, 3> unprom (nops);
unprom.quick_grow (nops);
unsigned int min_precision = 0;
bool single_use_p = false;
for (unsigned int i = 0; i < nops; ++i)
{
tree op = gimple_op (last_stmt, first_op + i);
if (TREE_CODE (op) == INTEGER_CST)
unprom[i].set_op (op, vect_constant_def);
else if (TREE_CODE (op) == SSA_NAME)
{
bool op_single_use_p = true;
if (!vect_look_through_possible_promotion (vinfo, op, &unprom[i],
&op_single_use_p))
return NULL;
/* If:
(1) N bits of the result are needed;
(2) all inputs are widened from M<N bits; and
(3) one operand OP is a single-use SSA name
we can shift the M->N widening from OP to the output
without changing the number or type of extensions involved.
This then reduces the number of copies of STMT_INFO.
If instead of (3) more than one operand is a single-use SSA name,
shifting the extension to the output is even more of a win.
If instead:
(1) N bits of the result are needed;
(2) one operand OP2 is widened from M2<N bits;
(3) another operand OP1 is widened from M1<M2 bits; and
(4) both OP1 and OP2 are single-use
the choice is between:
(a) truncating OP2 to M1, doing the operation on M1,
and then widening the result to N
(b) widening OP1 to M2, doing the operation on M2, and then
widening the result to N
Both shift the M2->N widening of the inputs to the output.
(a) additionally shifts the M1->M2 widening to the output;
it requires fewer copies of STMT_INFO but requires an extra
M2->M1 truncation.
Which is better will depend on the complexity and cost of
STMT_INFO, which is hard to predict at this stage. However,
a clear tie-breaker in favor of (b) is the fact that the
truncation in (a) increases the length of the operation chain.
If instead of (4) only one of OP1 or OP2 is single-use,
(b) is still a win over doing the operation in N bits:
it still shifts the M2->N widening on the single-use operand
to the output and reduces the number of STMT_INFO copies.
If neither operand is single-use then operating on fewer than
N bits might lead to more extensions overall. Whether it does
or not depends on global information about the vectorization
region, and whether that's a good trade-off would again
depend on the complexity and cost of the statements involved,
as well as things like register pressure that are not normally
modelled at this stage. We therefore ignore these cases
and just optimize the clear single-use wins above.
Thus we take the maximum precision of the unpromoted operands
and record whether any operand is single-use. */
if (unprom[i].dt == vect_internal_def)
{
min_precision = MAX (min_precision,
TYPE_PRECISION (unprom[i].type));
single_use_p |= op_single_use_p;
}
}
}
/* Although the operation could be done in operation_precision, we have
to balance that against introducing extra truncations or extensions.
Calculate the minimum precision that can be handled efficiently.
The loop above determined that the operation could be handled
efficiently in MIN_PRECISION if SINGLE_USE_P; this would shift an
extension from the inputs to the output without introducing more
instructions, and would reduce the number of instructions required
for STMT_INFO itself.
vect_determine_precisions has also determined that the result only
needs min_output_precision bits. Truncating by a factor of N times
requires a tree of N - 1 instructions, so if TYPE is N times wider
than min_output_precision, doing the operation in TYPE and truncating
the result requires N + (N - 1) = 2N - 1 instructions per output vector.
In contrast:
- truncating the input to a unary operation and doing the operation
in the new type requires at most N - 1 + 1 = N instructions per
output vector
- doing the same for a binary operation requires at most
(N - 1) * 2 + 1 = 2N - 1 instructions per output vector
Both unary and binary operations require fewer instructions than
this if the operands were extended from a suitable truncated form.
Thus there is usually nothing to lose by doing operations in
min_output_precision bits, but there can be something to gain. */
if (!single_use_p)
min_precision = last_stmt_info->min_output_precision;
else
min_precision = MIN (min_precision, last_stmt_info->min_output_precision);
/* Apply the minimum efficient precision we just calculated. */
if (new_precision < min_precision)
new_precision = min_precision;
new_precision = vect_element_precision (new_precision);
if (new_precision >= TYPE_PRECISION (type))
return NULL;
vect_pattern_detected ("vect_recog_over_widening_pattern", last_stmt);
*type_out = get_vectype_for_scalar_type (vinfo, type);
if (!*type_out)
return NULL;
/* We've found a viable pattern. Get the new type of the operation. */
bool unsigned_p = (last_stmt_info->operation_sign == UNSIGNED);
tree new_type = build_nonstandard_integer_type (new_precision, unsigned_p);
/* If we're truncating an operation, we need to make sure that we
don't introduce new undefined overflow. The codes tested here are
a subset of those accepted by vect_truncatable_operation_p. */
tree op_type = new_type;
if (TYPE_OVERFLOW_UNDEFINED (new_type)
&& (code == PLUS_EXPR || code == MINUS_EXPR || code == MULT_EXPR))
op_type = build_nonstandard_integer_type (new_precision, true);
/* We specifically don't check here whether the target supports the
new operation, since it might be something that a later pattern
wants to rewrite anyway. If targets have a minimum element size
for some optabs, we should pattern-match smaller ops to larger ops
where beneficial. */
tree new_vectype = get_vectype_for_scalar_type (vinfo, new_type);
tree op_vectype = get_vectype_for_scalar_type (vinfo, op_type);
if (!new_vectype || !op_vectype)
return NULL;
if (dump_enabled_p ())
dump_printf_loc (MSG_NOTE, vect_location, "demoting %T to %T\n",
type, new_type);
/* Calculate the rhs operands for an operation on OP_TYPE. */
tree ops[3] = {};
for (unsigned int i = 1; i < first_op; ++i)
ops[i - 1] = gimple_op (last_stmt, i);
vect_convert_inputs (last_stmt_info, nops, &ops[first_op - 1],
op_type, &unprom[0], op_vectype);
/* Use the operation to produce a result of type OP_TYPE. */
tree new_var = vect_recog_temp_ssa_var (op_type, NULL);
gimple *pattern_stmt = gimple_build_assign (new_var, code,
ops[0], ops[1], ops[2]);
gimple_set_location (pattern_stmt, gimple_location (last_stmt));
if (dump_enabled_p ())
dump_printf_loc (MSG_NOTE, vect_location,
"created pattern stmt: %G", pattern_stmt);
/* Convert back to the original signedness, if OP_TYPE is different
from NEW_TYPE. */
if (op_type != new_type)
pattern_stmt = vect_convert_output (last_stmt_info, new_type,
pattern_stmt, op_vectype);
/* Promote the result to the original type. */
pattern_stmt = vect_convert_output (last_stmt_info, type,
pattern_stmt, new_vectype);
return pattern_stmt;
}
/* Recognize the following patterns:
ATYPE a; // narrower than TYPE
BTYPE b; // narrower than TYPE
1) Multiply high with scaling
TYPE res = ((TYPE) a * (TYPE) b) >> c;
2) ... or also with rounding
TYPE res = (((TYPE) a * (TYPE) b) >> d + 1) >> 1;
where only the bottom half of res is used. */
static gimple *
vect_recog_mulhs_pattern (stmt_vec_info last_stmt_info, tree *type_out)
{
/* Check for a right shift. */
gassign *last_stmt = dyn_cast <gassign *> (last_stmt_info->stmt);
if (!last_stmt
|| gimple_assign_rhs_code (last_stmt) != RSHIFT_EXPR)
return NULL;
vec_info *vinfo = last_stmt_info->vinfo;
/* Check that the shift result is wider than the users of the
result need (i.e. that narrowing would be a natural choice). */
tree lhs_type = TREE_TYPE (gimple_assign_lhs (last_stmt));
unsigned int target_precision
= vect_element_precision (last_stmt_info->min_output_precision);
if (!INTEGRAL_TYPE_P (lhs_type)
|| target_precision >= TYPE_PRECISION (lhs_type))
return NULL;
/* Look through any change in sign on the outer shift input. */
vect_unpromoted_value unprom_rshift_input;
tree rshift_input = vect_look_through_possible_promotion
(vinfo, gimple_assign_rhs1 (last_stmt), &unprom_rshift_input);
if (!rshift_input
|| TYPE_PRECISION (TREE_TYPE (rshift_input))
!= TYPE_PRECISION (lhs_type))
return NULL;
/* Get the definition of the shift input. */
stmt_vec_info rshift_input_stmt_info
= vect_get_internal_def (vinfo, rshift_input);
if (!rshift_input_stmt_info)
return NULL;
gassign *rshift_input_stmt
= dyn_cast <gassign *> (rshift_input_stmt_info->stmt);
if (!rshift_input_stmt)
return NULL;
stmt_vec_info mulh_stmt_info;
tree scale_term;
internal_fn ifn;
unsigned int expect_offset;
/* Check for the presence of the rounding term. */
if (gimple_assign_rhs_code (rshift_input_stmt) == PLUS_EXPR)
{
/* Check that the outer shift was by 1. */
if (!integer_onep (gimple_assign_rhs2 (last_stmt)))
return NULL;
/* Check that the second operand of the PLUS_EXPR is 1. */
if (!integer_onep (gimple_assign_rhs2 (rshift_input_stmt)))
return NULL;
/* Look through any change in sign on the addition input. */
vect_unpromoted_value unprom_plus_input;
tree plus_input = vect_look_through_possible_promotion
(vinfo, gimple_assign_rhs1 (rshift_input_stmt), &unprom_plus_input);
if (!plus_input
|| TYPE_PRECISION (TREE_TYPE (plus_input))
!= TYPE_PRECISION (TREE_TYPE (rshift_input)))
return NULL;
/* Get the definition of the multiply-high-scale part. */
stmt_vec_info plus_input_stmt_info
= vect_get_internal_def (vinfo, plus_input);
if (!plus_input_stmt_info)
return NULL;
gassign *plus_input_stmt
= dyn_cast <gassign *> (plus_input_stmt_info->stmt);
if (!plus_input_stmt
|| gimple_assign_rhs_code (plus_input_stmt) != RSHIFT_EXPR)
return NULL;
/* Look through any change in sign on the scaling input. */
vect_unpromoted_value unprom_scale_input;
tree scale_input = vect_look_through_possible_promotion
(vinfo, gimple_assign_rhs1 (plus_input_stmt), &unprom_scale_input);
if (!scale_input
|| TYPE_PRECISION (TREE_TYPE (scale_input))
!= TYPE_PRECISION (TREE_TYPE (plus_input)))
return NULL;
/* Get the definition of the multiply-high part. */
mulh_stmt_info = vect_get_internal_def (vinfo, scale_input);
if (!mulh_stmt_info)
return NULL;
/* Get the scaling term. */
scale_term = gimple_assign_rhs2 (plus_input_stmt);
expect_offset = target_precision + 2;
ifn = IFN_MULHRS;
}
else
{
mulh_stmt_info = rshift_input_stmt_info;
scale_term = gimple_assign_rhs2 (last_stmt);
expect_offset = target_precision + 1;
ifn = IFN_MULHS;
}
/* Check that the scaling factor is correct. */
if (TREE_CODE (scale_term) != INTEGER_CST
|| wi::to_widest (scale_term) + expect_offset
!= TYPE_PRECISION (lhs_type))
return NULL;
/* Check whether the scaling input term can be seen as two widened
inputs multiplied together. */
vect_unpromoted_value unprom_mult[2];
tree new_type;
unsigned int nops
= vect_widened_op_tree (mulh_stmt_info, MULT_EXPR, WIDEN_MULT_EXPR,
false, 2, unprom_mult, &new_type);
if (nops != 2)
return NULL;
vect_pattern_detected ("vect_recog_mulhs_pattern", last_stmt);
/* Adjust output precision. */
if (TYPE_PRECISION (new_type) < target_precision)
new_type = build_nonstandard_integer_type
(target_precision, TYPE_UNSIGNED (new_type));
/* Check for target support. */
tree new_vectype = get_vectype_for_scalar_type (vinfo, new_type);
if (!new_vectype
|| !direct_internal_fn_supported_p
(ifn, new_vectype, OPTIMIZE_FOR_SPEED))
return NULL;
/* The IR requires a valid vector type for the cast result, even though
it's likely to be discarded. */
*type_out = get_vectype_for_scalar_type (vinfo, lhs_type);
if (!*type_out)
return NULL;
/* Generate the IFN_MULHRS call. */
tree new_var = vect_recog_temp_ssa_var (new_type, NULL);
tree new_ops[2];
vect_convert_inputs (last_stmt_info, 2, new_ops, new_type,
unprom_mult, new_vectype);
gcall *mulhrs_stmt
= gimple_build_call_internal (ifn, 2, new_ops[0], new_ops[1]);
gimple_call_set_lhs (mulhrs_stmt, new_var);
gimple_set_location (mulhrs_stmt, gimple_location (last_stmt));
if (dump_enabled_p ())
dump_printf_loc (MSG_NOTE, vect_location,
"created pattern stmt: %G", mulhrs_stmt);
return vect_convert_output (last_stmt_info, lhs_type,
mulhrs_stmt, new_vectype);
}
/* Recognize the patterns:
ATYPE a; // narrower than TYPE
BTYPE b; // narrower than TYPE
(1) TYPE avg = ((TYPE) a + (TYPE) b) >> 1;
or (2) TYPE avg = ((TYPE) a + (TYPE) b + 1) >> 1;
where only the bottom half of avg is used. Try to transform them into:
(1) NTYPE avg' = .AVG_FLOOR ((NTYPE) a, (NTYPE) b);
or (2) NTYPE avg' = .AVG_CEIL ((NTYPE) a, (NTYPE) b);
followed by:
TYPE avg = (TYPE) avg';
where NTYPE is no wider than half of TYPE. Since only the bottom half
of avg is used, all or part of the cast of avg' should become redundant.
If there is no target support available, generate code to distribute rshift
over plus and add a carry. */
static gimple *
vect_recog_average_pattern (stmt_vec_info last_stmt_info, tree *type_out)
{
/* Check for a shift right by one bit. */
gassign *last_stmt = dyn_cast <gassign *> (last_stmt_info->stmt);
vec_info *vinfo = last_stmt_info->vinfo;
if (!last_stmt
|| gimple_assign_rhs_code (last_stmt) != RSHIFT_EXPR
|| !integer_onep (gimple_assign_rhs2 (last_stmt)))
return NULL;
/* Check that the shift result is wider than the users of the
result need (i.e. that narrowing would be a natural choice). */
tree lhs = gimple_assign_lhs (last_stmt);
tree type = TREE_TYPE (lhs);
unsigned int target_precision
= vect_element_precision (last_stmt_info->min_output_precision);
if (!INTEGRAL_TYPE_P (type) || target_precision >= TYPE_PRECISION (type))
return NULL;
/* Look through any change in sign on the shift input. */
tree rshift_rhs = gimple_assign_rhs1 (last_stmt);
vect_unpromoted_value unprom_plus;
rshift_rhs = vect_look_through_possible_promotion (vinfo, rshift_rhs,
&unprom_plus);
if (!rshift_rhs
|| TYPE_PRECISION (TREE_TYPE (rshift_rhs)) != TYPE_PRECISION (type))
return NULL;
/* Get the definition of the shift input. */
stmt_vec_info plus_stmt_info = vect_get_internal_def (vinfo, rshift_rhs);
if (!plus_stmt_info)
return NULL;
/* Check whether the shift input can be seen as a tree of additions on
2 or 3 widened inputs.
Note that the pattern should be a win even if the result of one or
more additions is reused elsewhere: if the pattern matches, we'd be
replacing 2N RSHIFT_EXPRs and N VEC_PACK_*s with N IFN_AVG_*s. */
internal_fn ifn = IFN_AVG_FLOOR;
vect_unpromoted_value unprom[3];
tree new_type;
unsigned int nops = vect_widened_op_tree (plus_stmt_info, PLUS_EXPR,
PLUS_EXPR, false, 3,
unprom, &new_type);
if (nops == 0)
return NULL;
if (nops == 3)
{
/* Check that one operand is 1. */
unsigned int i;
for (i = 0; i < 3; ++i)
if (integer_onep (unprom[i].op))
break;
if (i == 3)
return NULL;
/* Throw away the 1 operand and keep the other two. */
if (i < 2)
unprom[i] = unprom[2];
ifn = IFN_AVG_CEIL;
}
vect_pattern_detected ("vect_recog_average_pattern", last_stmt);
/* We know that:
(a) the operation can be viewed as:
TYPE widened0 = (TYPE) UNPROM[0];
TYPE widened1 = (TYPE) UNPROM[1];
TYPE tmp1 = widened0 + widened1 {+ 1};
TYPE tmp2 = tmp1 >> 1; // LAST_STMT_INFO
(b) the first two statements are equivalent to:
TYPE widened0 = (TYPE) (NEW_TYPE) UNPROM[0];
TYPE widened1 = (TYPE) (NEW_TYPE) UNPROM[1];
(c) vect_recog_over_widening_pattern has already tried to narrow TYPE
where sensible;
(d) all the operations can be performed correctly at twice the width of
NEW_TYPE, due to the nature of the average operation; and
(e) users of the result of the right shift need only TARGET_PRECISION
bits, where TARGET_PRECISION is no more than half of TYPE's
precision.
Under these circumstances, the only situation in which NEW_TYPE
could be narrower than TARGET_PRECISION is if widened0, widened1
and an addition result are all used more than once. Thus we can
treat any widening of UNPROM[0] and UNPROM[1] to TARGET_PRECISION
as "free", whereas widening the result of the average instruction
from NEW_TYPE to TARGET_PRECISION would be a new operation. It's
therefore better not to go narrower than TARGET_PRECISION. */
if (TYPE_PRECISION (new_type) < target_precision)
new_type = build_nonstandard_integer_type (target_precision,
TYPE_UNSIGNED (new_type));
/* Check for target support. */
tree new_vectype = get_vectype_for_scalar_type (vinfo, new_type);
if (!new_vectype)
return NULL;
bool fallback_p = false;
if (direct_internal_fn_supported_p (ifn, new_vectype, OPTIMIZE_FOR_SPEED))
;
else if (TYPE_UNSIGNED (new_type)
&& optab_for_tree_code (RSHIFT_EXPR, new_vectype, optab_scalar)
&& optab_for_tree_code (PLUS_EXPR, new_vectype, optab_default)
&& optab_for_tree_code (BIT_IOR_EXPR, new_vectype, optab_default)
&& optab_for_tree_code (BIT_AND_EXPR, new_vectype, optab_default))
fallback_p = true;
else
return NULL;
/* The IR requires a valid vector type for the cast result, even though
it's likely to be discarded. */
*type_out = get_vectype_for_scalar_type (vinfo, type);
if (!*type_out)
return NULL;
tree new_var = vect_recog_temp_ssa_var (new_type, NULL);
tree new_ops[2];
vect_convert_inputs (last_stmt_info, 2, new_ops, new_type,
unprom, new_vectype);
if (fallback_p)
{
/* As a fallback, generate code for following sequence:
shifted_op0 = new_ops[0] >> 1;
shifted_op1 = new_ops[1] >> 1;
sum_of_shifted = shifted_op0 + shifted_op1;
unmasked_carry = new_ops[0] and/or new_ops[1];
carry = unmasked_carry & 1;
new_var = sum_of_shifted + carry;
*/
tree one_cst = build_one_cst (new_type);
gassign *g;
tree shifted_op0 = vect_recog_temp_ssa_var (new_type, NULL);
g = gimple_build_assign (shifted_op0, RSHIFT_EXPR, new_ops[0], one_cst);
append_pattern_def_seq (last_stmt_info, g, new_vectype);
tree shifted_op1 = vect_recog_temp_ssa_var (new_type, NULL);
g = gimple_build_assign (shifted_op1, RSHIFT_EXPR, new_ops[1], one_cst);
append_pattern_def_seq (last_stmt_info, g, new_vectype);
tree sum_of_shifted = vect_recog_temp_ssa_var (new_type, NULL);
g = gimple_build_assign (sum_of_shifted, PLUS_EXPR,
shifted_op0, shifted_op1);
append_pattern_def_seq (last_stmt_info, g, new_vectype);
tree unmasked_carry = vect_recog_temp_ssa_var (new_type, NULL);
tree_code c = (ifn == IFN_AVG_CEIL) ? BIT_IOR_EXPR : BIT_AND_EXPR;
g = gimple_build_assign (unmasked_carry, c, new_ops[0], new_ops[1]);
append_pattern_def_seq (last_stmt_info, g, new_vectype);
tree carry = vect_recog_temp_ssa_var (new_type, NULL);
g = gimple_build_assign (carry, BIT_AND_EXPR, unmasked_carry, one_cst);
append_pattern_def_seq (last_stmt_info, g, new_vectype);
g = gimple_build_assign (new_var, PLUS_EXPR, sum_of_shifted, carry);
return vect_convert_output (last_stmt_info, type, g, new_vectype);
}
/* Generate the IFN_AVG* call. */
gcall *average_stmt = gimple_build_call_internal (ifn, 2, new_ops[0],
new_ops[1]);
gimple_call_set_lhs (average_stmt, new_var);
gimple_set_location (average_stmt, gimple_location (last_stmt));
if (dump_enabled_p ())
dump_printf_loc (MSG_NOTE, vect_location,
"created pattern stmt: %G", average_stmt);
return vect_convert_output (last_stmt_info, type, average_stmt, new_vectype);
}
/* Recognize cases in which the input to a cast is wider than its
output, and the input is fed by a widening operation. Fold this
by removing the unnecessary intermediate widening. E.g.:
unsigned char a;
unsigned int b = (unsigned int) a;
unsigned short c = (unsigned short) b;
-->
unsigned short c = (unsigned short) a;
Although this is rare in input IR, it is an expected side-effect
of the over-widening pattern above.
This is beneficial also for integer-to-float conversions, if the
widened integer has more bits than the float, and if the unwidened
input doesn't. */
static gimple *
vect_recog_cast_forwprop_pattern (stmt_vec_info last_stmt_info, tree *type_out)
{
/* Check for a cast, including an integer-to-float conversion. */
gassign *last_stmt = dyn_cast <gassign *> (last_stmt_info->stmt);
if (!last_stmt)
return NULL;
tree_code code = gimple_assign_rhs_code (last_stmt);
if (!CONVERT_EXPR_CODE_P (code) && code != FLOAT_EXPR)
return NULL;
/* Make sure that the rhs is a scalar with a natural bitsize. */
tree lhs = gimple_assign_lhs (last_stmt);
if (!lhs)
return NULL;
tree lhs_type = TREE_TYPE (lhs);
scalar_mode lhs_mode;
if (VECT_SCALAR_BOOLEAN_TYPE_P (lhs_type)
|| !is_a <scalar_mode> (TYPE_MODE (lhs_type), &lhs_mode))
return NULL;
/* Check for a narrowing operation (from a vector point of view). */
tree rhs = gimple_assign_rhs1 (last_stmt);
tree rhs_type = TREE_TYPE (rhs);
if (!INTEGRAL_TYPE_P (rhs_type)
|| VECT_SCALAR_BOOLEAN_TYPE_P (rhs_type)
|| TYPE_PRECISION (rhs_type) <= GET_MODE_BITSIZE (lhs_mode))
return NULL;
/* Try to find an unpromoted input. */
vec_info *vinfo = last_stmt_info->vinfo;
vect_unpromoted_value unprom;
if (!vect_look_through_possible_promotion (vinfo, rhs, &unprom)
|| TYPE_PRECISION (unprom.type) >= TYPE_PRECISION (rhs_type))
return NULL;
/* If the bits above RHS_TYPE matter, make sure that they're the
same when extending from UNPROM as they are when extending from RHS. */
if (!INTEGRAL_TYPE_P (lhs_type)
&& TYPE_SIGN (rhs_type) != TYPE_SIGN (unprom.type))
return NULL;
/* We can get the same result by casting UNPROM directly, to avoid
the unnecessary widening and narrowing. */
vect_pattern_detected ("vect_recog_cast_forwprop_pattern", last_stmt);
*type_out = get_vectype_for_scalar_type (vinfo, lhs_type);
if (!*type_out)
return NULL;
tree new_var = vect_recog_temp_ssa_var (lhs_type, NULL);
gimple *pattern_stmt = gimple_build_assign (new_var, code, unprom.op);
gimple_set_location (pattern_stmt, gimple_location (last_stmt));
return pattern_stmt;
}
/* Try to detect a shift left of a widened input, converting LSHIFT_EXPR
to WIDEN_LSHIFT_EXPR. See vect_recog_widen_op_pattern for details. */
static gimple *
vect_recog_widen_shift_pattern (stmt_vec_info last_stmt_info, tree *type_out)
{
return vect_recog_widen_op_pattern (last_stmt_info, type_out, LSHIFT_EXPR,
WIDEN_LSHIFT_EXPR, true,
"vect_recog_widen_shift_pattern");
}
/* Detect a rotate pattern wouldn't be otherwise vectorized:
type a_t, b_t, c_t;
S0 a_t = b_t r<< c_t;
Input/Output:
* STMT_VINFO: The stmt from which the pattern search begins,
i.e. the shift/rotate stmt. The original stmt (S0) is replaced
with a sequence:
S1 d_t = -c_t;
S2 e_t = d_t & (B - 1);
S3 f_t = b_t << c_t;
S4 g_t = b_t >> e_t;
S0 a_t = f_t | g_t;
where B is element bitsize of type.
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace the rotate
S0 stmt. */
static gimple *
vect_recog_rotate_pattern (stmt_vec_info stmt_vinfo, tree *type_out)
{
gimple *last_stmt = stmt_vinfo->stmt;
tree oprnd0, oprnd1, lhs, var, var1, var2, vectype, type, stype, def, def2;
gimple *pattern_stmt, *def_stmt;
enum tree_code rhs_code;
vec_info *vinfo = stmt_vinfo->vinfo;
enum vect_def_type dt;
optab optab1, optab2;
edge ext_def = NULL;
bool bswap16_p = false;
if (is_gimple_assign (last_stmt))
{
rhs_code = gimple_assign_rhs_code (last_stmt);
switch (rhs_code)
{
case LROTATE_EXPR:
case RROTATE_EXPR:
break;
default:
return NULL;
}
lhs = gimple_assign_lhs (last_stmt);
oprnd0 = gimple_assign_rhs1 (last_stmt);
type = TREE_TYPE (oprnd0);
oprnd1 = gimple_assign_rhs2 (last_stmt);
}
else if (gimple_call_builtin_p (last_stmt, BUILT_IN_BSWAP16))
{
/* __builtin_bswap16 (x) is another form of x r>> 8.
The vectorizer has bswap support, but only if the argument isn't
promoted. */
lhs = gimple_call_lhs (last_stmt);
oprnd0 = gimple_call_arg (last_stmt, 0);
type = TREE_TYPE (oprnd0);
if (!lhs
|| TYPE_PRECISION (TREE_TYPE (lhs)) != 16
|| TYPE_PRECISION (type) <= 16
|| TREE_CODE (oprnd0) != SSA_NAME
|| BITS_PER_UNIT != 8
|| !TYPE_UNSIGNED (TREE_TYPE (lhs)))
return NULL;
stmt_vec_info def_stmt_info;
if (!vect_is_simple_use (oprnd0, vinfo, &dt, &def_stmt_info, &def_stmt))
return NULL;
if (dt != vect_internal_def)
return NULL;
if (gimple_assign_cast_p (def_stmt))
{
def = gimple_assign_rhs1 (def_stmt);
if (INTEGRAL_TYPE_P (TREE_TYPE (def))
&& TYPE_PRECISION (TREE_TYPE (def)) == 16)
oprnd0 = def;
}
type = TREE_TYPE (lhs);
vectype = get_vectype_for_scalar_type (vinfo, type);
if (vectype == NULL_TREE)
return NULL;
if (tree char_vectype = get_same_sized_vectype (char_type_node, vectype))
{
/* The encoding uses one stepped pattern for each byte in the
16-bit word. */
vec_perm_builder elts (TYPE_VECTOR_SUBPARTS (char_vectype), 2, 3);
for (unsigned i = 0; i < 3; ++i)
for (unsigned j = 0; j < 2; ++j)
elts.quick_push ((i + 1) * 2 - j - 1);
vec_perm_indices indices (elts, 1,
TYPE_VECTOR_SUBPARTS (char_vectype));
if (can_vec_perm_const_p (TYPE_MODE (char_vectype), indices))
{
/* vectorizable_bswap can handle the __builtin_bswap16 if we
undo the argument promotion. */
if (!useless_type_conversion_p (type, TREE_TYPE (oprnd0)))
{
def = vect_recog_temp_ssa_var (type, NULL);
def_stmt = gimple_build_assign (def, NOP_EXPR, oprnd0);
append_pattern_def_seq (stmt_vinfo, def_stmt);
oprnd0 = def;
}
/* Pattern detected. */
vect_pattern_detected ("vect_recog_rotate_pattern", last_stmt);
*type_out = vectype;
/* Pattern supported. Create a stmt to be used to replace the
pattern, with the unpromoted argument. */
var = vect_recog_temp_ssa_var (type, NULL);
pattern_stmt = gimple_build_call (gimple_call_fndecl (last_stmt),
1, oprnd0);
gimple_call_set_lhs (pattern_stmt, var);
gimple_call_set_fntype (as_a <gcall *> (pattern_stmt),
gimple_call_fntype (last_stmt));
return pattern_stmt;
}
}
oprnd1 = build_int_cst (integer_type_node, 8);
rhs_code = LROTATE_EXPR;
bswap16_p = true;
}
else
return NULL;
if (TREE_CODE (oprnd0) != SSA_NAME
|| TYPE_PRECISION (TREE_TYPE (lhs)) != TYPE_PRECISION (type)
|| !INTEGRAL_TYPE_P (type)
|| !TYPE_UNSIGNED (type))
return NULL;
stmt_vec_info def_stmt_info;
if (!vect_is_simple_use (oprnd1, vinfo, &dt, &def_stmt_info, &def_stmt))
return NULL;
if (dt != vect_internal_def
&& dt != vect_constant_def
&& dt != vect_external_def)
return NULL;
vectype = get_vectype_for_scalar_type (vinfo, type);
if (vectype == NULL_TREE)
return NULL;
/* If vector/vector or vector/scalar rotate is supported by the target,
don't do anything here. */
optab1 = optab_for_tree_code (rhs_code, vectype, optab_vector);
if (optab1
&& optab_handler (optab1, TYPE_MODE (vectype)) != CODE_FOR_nothing)
{
use_rotate:
if (bswap16_p)
{
if (!useless_type_conversion_p (type, TREE_TYPE (oprnd0)))
{
def = vect_recog_temp_ssa_var (type, NULL);
def_stmt = gimple_build_assign (def, NOP_EXPR, oprnd0);
append_pattern_def_seq (stmt_vinfo, def_stmt);
oprnd0 = def;
}
/* Pattern detected. */
vect_pattern_detected ("vect_recog_rotate_pattern", last_stmt);
*type_out = vectype;
/* Pattern supported. Create a stmt to be used to replace the
pattern. */
var = vect_recog_temp_ssa_var (type, NULL);
pattern_stmt = gimple_build_assign (var, LROTATE_EXPR, oprnd0,
oprnd1);
return pattern_stmt;
}
return NULL;
}
if (is_a <bb_vec_info> (vinfo) || dt != vect_internal_def)
{
optab2 = optab_for_tree_code (rhs_code, vectype, optab_scalar);
if (optab2
&& optab_handler (optab2, TYPE_MODE (vectype)) != CODE_FOR_nothing)
goto use_rotate;
}
/* If vector/vector or vector/scalar shifts aren't supported by the target,
don't do anything here either. */
optab1 = optab_for_tree_code (LSHIFT_EXPR, vectype, optab_vector);
optab2 = optab_for_tree_code (RSHIFT_EXPR, vectype, optab_vector);
if (!optab1
|| optab_handler (optab1, TYPE_MODE (vectype)) == CODE_FOR_nothing
|| !optab2
|| optab_handler (optab2, TYPE_MODE (vectype)) == CODE_FOR_nothing)
{
if (! is_a <bb_vec_info> (vinfo) && dt == vect_internal_def)
return NULL;
optab1 = optab_for_tree_code (LSHIFT_EXPR, vectype, optab_scalar);
optab2 = optab_for_tree_code (RSHIFT_EXPR, vectype, optab_scalar);
if (!optab1
|| optab_handler (optab1, TYPE_MODE (vectype)) == CODE_FOR_nothing
|| !optab2
|| optab_handler (optab2, TYPE_MODE (vectype)) == CODE_FOR_nothing)
return NULL;
}
*type_out = vectype;
if (bswap16_p && !useless_type_conversion_p (type, TREE_TYPE (oprnd0)))
{
def = vect_recog_temp_ssa_var (type, NULL);
def_stmt = gimple_build_assign (def, NOP_EXPR, oprnd0);
append_pattern_def_seq (stmt_vinfo, def_stmt);
oprnd0 = def;
}
if (dt == vect_external_def && TREE_CODE (oprnd1) == SSA_NAME)
ext_def = vect_get_external_def_edge (vinfo, oprnd1);
def = NULL_TREE;
scalar_int_mode mode = SCALAR_INT_TYPE_MODE (type);
if (dt != vect_internal_def || TYPE_MODE (TREE_TYPE (oprnd1)) == mode)
def = oprnd1;
else if (def_stmt && gimple_assign_cast_p (def_stmt))
{
tree rhs1 = gimple_assign_rhs1 (def_stmt);
if (TYPE_MODE (TREE_TYPE (rhs1)) == mode
&& TYPE_PRECISION (TREE_TYPE (rhs1))
== TYPE_PRECISION (type))
def = rhs1;
}
if (def == NULL_TREE)
{
def = vect_recog_temp_ssa_var (type, NULL);
def_stmt = gimple_build_assign (def, NOP_EXPR, oprnd1);
append_pattern_def_seq (stmt_vinfo, def_stmt);
}
stype = TREE_TYPE (def);
if (TREE_CODE (def) == INTEGER_CST)
{
if (!tree_fits_uhwi_p (def)
|| tree_to_uhwi (def) >= GET_MODE_PRECISION (mode)
|| integer_zerop (def))
return NULL;
def2 = build_int_cst (stype,
GET_MODE_PRECISION (mode) - tree_to_uhwi (def));
}
else
{
tree vecstype = get_vectype_for_scalar_type (vinfo, stype);
if (vecstype == NULL_TREE)
return NULL;
def2 = vect_recog_temp_ssa_var (stype, NULL);
def_stmt = gimple_build_assign (def2, NEGATE_EXPR, def);
if (ext_def)
{
basic_block new_bb
= gsi_insert_on_edge_immediate (ext_def, def_stmt);
gcc_assert (!new_bb);
}
else
append_pattern_def_seq (stmt_vinfo, def_stmt, vecstype);
def2 = vect_recog_temp_ssa_var (stype, NULL);
tree mask = build_int_cst (stype, GET_MODE_PRECISION (mode) - 1);
def_stmt = gimple_build_assign (def2, BIT_AND_EXPR,
gimple_assign_lhs (def_stmt), mask);
if (ext_def)
{
basic_block new_bb
= gsi_insert_on_edge_immediate (ext_def, def_stmt);
gcc_assert (!new_bb);
}
else
append_pattern_def_seq (stmt_vinfo, def_stmt, vecstype);
}
var1 = vect_recog_temp_ssa_var (type, NULL);
def_stmt = gimple_build_assign (var1, rhs_code == LROTATE_EXPR
? LSHIFT_EXPR : RSHIFT_EXPR,
oprnd0, def);
append_pattern_def_seq (stmt_vinfo, def_stmt);
var2 = vect_recog_temp_ssa_var (type, NULL);
def_stmt = gimple_build_assign (var2, rhs_code == LROTATE_EXPR
? RSHIFT_EXPR : LSHIFT_EXPR,
oprnd0, def2);
append_pattern_def_seq (stmt_vinfo, def_stmt);
/* Pattern detected. */
vect_pattern_detected ("vect_recog_rotate_pattern", last_stmt);
/* Pattern supported. Create a stmt to be used to replace the pattern. */
var = vect_recog_temp_ssa_var (type, NULL);
pattern_stmt = gimple_build_assign (var, BIT_IOR_EXPR, var1, var2);
return pattern_stmt;
}
/* Detect a vector by vector shift pattern that wouldn't be otherwise
vectorized:
type a_t;
TYPE b_T, res_T;
S1 a_t = ;
S2 b_T = ;
S3 res_T = b_T op a_t;
where type 'TYPE' is a type with different size than 'type',
and op is <<, >> or rotate.
Also detect cases:
type a_t;
TYPE b_T, c_T, res_T;
S0 c_T = ;
S1 a_t = (type) c_T;
S2 b_T = ;
S3 res_T = b_T op a_t;
Input/Output:
* STMT_VINFO: The stmt from which the pattern search begins,
i.e. the shift/rotate stmt. The original stmt (S3) is replaced
with a shift/rotate which has same type on both operands, in the
second case just b_T op c_T, in the first case with added cast
from a_t to c_T in STMT_VINFO_PATTERN_DEF_SEQ.
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace the shift/rotate
S3 stmt. */
static gimple *
vect_recog_vector_vector_shift_pattern (stmt_vec_info stmt_vinfo,
tree *type_out)
{
gimple *last_stmt = stmt_vinfo->stmt;
tree oprnd0, oprnd1, lhs, var;
gimple *pattern_stmt;
enum tree_code rhs_code;
vec_info *vinfo = stmt_vinfo->vinfo;
if (!is_gimple_assign (last_stmt))
return NULL;
rhs_code = gimple_assign_rhs_code (last_stmt);
switch (rhs_code)
{
case LSHIFT_EXPR:
case RSHIFT_EXPR:
case LROTATE_EXPR:
case RROTATE_EXPR:
break;
default:
return NULL;
}
lhs = gimple_assign_lhs (last_stmt);
oprnd0 = gimple_assign_rhs1 (last_stmt);
oprnd1 = gimple_assign_rhs2 (last_stmt);
if (TREE_CODE (oprnd0) != SSA_NAME
|| TREE_CODE (oprnd1) != SSA_NAME
|| TYPE_MODE (TREE_TYPE (oprnd0)) == TYPE_MODE (TREE_TYPE (oprnd1))
|| !type_has_mode_precision_p (TREE_TYPE (oprnd1))
|| TYPE_PRECISION (TREE_TYPE (lhs))
!= TYPE_PRECISION (TREE_TYPE (oprnd0)))
return NULL;
stmt_vec_info def_vinfo = vect_get_internal_def (vinfo, oprnd1);
if (!def_vinfo)
return NULL;
*type_out = get_vectype_for_scalar_type (vinfo, TREE_TYPE (oprnd0));
if (*type_out == NULL_TREE)
return NULL;
tree def = NULL_TREE;
gassign *def_stmt = dyn_cast <gassign *> (def_vinfo->stmt);
if (def_stmt && gimple_assign_cast_p (def_stmt))
{
tree rhs1 = gimple_assign_rhs1 (def_stmt);
if (TYPE_MODE (TREE_TYPE (rhs1)) == TYPE_MODE (TREE_TYPE (oprnd0))
&& TYPE_PRECISION (TREE_TYPE (rhs1))
== TYPE_PRECISION (TREE_TYPE (oprnd0)))
{
if (TYPE_PRECISION (TREE_TYPE (oprnd1))
>= TYPE_PRECISION (TREE_TYPE (rhs1)))
def = rhs1;
else
{
tree mask
= build_low_bits_mask (TREE_TYPE (rhs1),
TYPE_PRECISION (TREE_TYPE (oprnd1)));
def = vect_recog_temp_ssa_var (TREE_TYPE (rhs1), NULL);
def_stmt = gimple_build_assign (def, BIT_AND_EXPR, rhs1, mask);
tree vecstype = get_vectype_for_scalar_type (vinfo,
TREE_TYPE (rhs1));
append_pattern_def_seq (stmt_vinfo, def_stmt, vecstype);
}
}
}
if (def == NULL_TREE)
{
def = vect_recog_temp_ssa_var (TREE_TYPE (oprnd0), NULL);
def_stmt = gimple_build_assign (def, NOP_EXPR, oprnd1);
append_pattern_def_seq (stmt_vinfo, def_stmt);
}
/* Pattern detected. */
vect_pattern_detected ("vect_recog_vector_vector_shift_pattern", last_stmt);
/* Pattern supported. Create a stmt to be used to replace the pattern. */
var = vect_recog_temp_ssa_var (TREE_TYPE (oprnd0), NULL);
pattern_stmt = gimple_build_assign (var, rhs_code, oprnd0, def);
return pattern_stmt;
}
/* Return true iff the target has a vector optab implementing the operation
CODE on type VECTYPE. */
static bool
target_has_vecop_for_code (tree_code code, tree vectype)
{
optab voptab = optab_for_tree_code (code, vectype, optab_vector);
return voptab
&& optab_handler (voptab, TYPE_MODE (vectype)) != CODE_FOR_nothing;
}
/* Verify that the target has optabs of VECTYPE to perform all the steps
needed by the multiplication-by-immediate synthesis algorithm described by
ALG and VAR. If SYNTH_SHIFT_P is true ensure that vector addition is
present. Return true iff the target supports all the steps. */
static bool
target_supports_mult_synth_alg (struct algorithm *alg, mult_variant var,
tree vectype, bool synth_shift_p)
{
if (alg->op[0] != alg_zero && alg->op[0] != alg_m)
return false;
bool supports_vminus = target_has_vecop_for_code (MINUS_EXPR, vectype);
bool supports_vplus = target_has_vecop_for_code (PLUS_EXPR, vectype);
if (var == negate_variant
&& !target_has_vecop_for_code (NEGATE_EXPR, vectype))
return false;
/* If we must synthesize shifts with additions make sure that vector
addition is available. */
if ((var == add_variant || synth_shift_p) && !supports_vplus)
return false;
for (int i = 1; i < alg->ops; i++)
{
switch (alg->op[i])
{
case alg_shift:
break;
case alg_add_t_m2:
case alg_add_t2_m:
case alg_add_factor:
if (!supports_vplus)
return false;
break;
case alg_sub_t_m2:
case alg_sub_t2_m:
case alg_sub_factor:
if (!supports_vminus)
return false;
break;
case alg_unknown:
case alg_m:
case alg_zero:
case alg_impossible:
return false;
default:
gcc_unreachable ();
}
}
return true;
}
/* Synthesize a left shift of OP by AMNT bits using a series of additions and
putting the final result in DEST. Append all statements but the last into
VINFO. Return the last statement. */
static gimple *
synth_lshift_by_additions (tree dest, tree op, HOST_WIDE_INT amnt,
stmt_vec_info vinfo)
{
HOST_WIDE_INT i;
tree itype = TREE_TYPE (op);
tree prev_res = op;
gcc_assert (amnt >= 0);
for (i = 0; i < amnt; i++)
{
tree tmp_var = (i < amnt - 1) ? vect_recog_temp_ssa_var (itype, NULL)
: dest;
gimple *stmt
= gimple_build_assign (tmp_var, PLUS_EXPR, prev_res, prev_res);
prev_res = tmp_var;
if (i < amnt - 1)
append_pattern_def_seq (vinfo, stmt);
else
return stmt;
}
gcc_unreachable ();
return NULL;
}
/* Helper for vect_synth_mult_by_constant. Apply a binary operation
CODE to operands OP1 and OP2, creating a new temporary SSA var in
the process if necessary. Append the resulting assignment statements
to the sequence in STMT_VINFO. Return the SSA variable that holds the
result of the binary operation. If SYNTH_SHIFT_P is true synthesize
left shifts using additions. */
static tree
apply_binop_and_append_stmt (tree_code code, tree op1, tree op2,
stmt_vec_info stmt_vinfo, bool synth_shift_p)
{
if (integer_zerop (op2)
&& (code == LSHIFT_EXPR
|| code == PLUS_EXPR))
{
gcc_assert (TREE_CODE (op1) == SSA_NAME);
return op1;
}
gimple *stmt;
tree itype = TREE_TYPE (op1);
tree tmp_var = vect_recog_temp_ssa_var (itype, NULL);
if (code == LSHIFT_EXPR
&& synth_shift_p)
{
stmt = synth_lshift_by_additions (tmp_var, op1, TREE_INT_CST_LOW (op2),
stmt_vinfo);
append_pattern_def_seq (stmt_vinfo, stmt);
return tmp_var;
}
stmt = gimple_build_assign (tmp_var, code, op1, op2);
append_pattern_def_seq (stmt_vinfo, stmt);
return tmp_var;
}
/* Synthesize a multiplication of OP by an INTEGER_CST VAL using shifts
and simple arithmetic operations to be vectorized. Record the statements
produced in STMT_VINFO and return the last statement in the sequence or
NULL if it's not possible to synthesize such a multiplication.
This function mirrors the behavior of expand_mult_const in expmed.c but
works on tree-ssa form. */
static gimple *
vect_synth_mult_by_constant (tree op, tree val,
stmt_vec_info stmt_vinfo)
{
vec_info *vinfo = stmt_vinfo->vinfo;
tree itype = TREE_TYPE (op);
machine_mode mode = TYPE_MODE (itype);
struct algorithm alg;
mult_variant variant;
if (!tree_fits_shwi_p (val))
return NULL;
/* Multiplication synthesis by shifts, adds and subs can introduce
signed overflow where the original operation didn't. Perform the
operations on an unsigned type and cast back to avoid this.
In the future we may want to relax this for synthesis algorithms
that we can prove do not cause unexpected overflow. */
bool cast_to_unsigned_p = !TYPE_OVERFLOW_WRAPS (itype);
tree multtype = cast_to_unsigned_p ? unsigned_type_for (itype) : itype;
/* Targets that don't support vector shifts but support vector additions
can synthesize shifts that way. */
bool synth_shift_p = !vect_supportable_shift (vinfo, LSHIFT_EXPR, multtype);
HOST_WIDE_INT hwval = tree_to_shwi (val);
/* Use MAX_COST here as we don't want to limit the sequence on rtx costs.
The vectorizer's benefit analysis will decide whether it's beneficial
to do this. */
bool possible = choose_mult_variant (mode, hwval, &alg,
&variant, MAX_COST);
if (!possible)
return NULL;
tree vectype = get_vectype_for_scalar_type (vinfo, multtype);
if (!vectype
|| !target_supports_mult_synth_alg (&alg, variant,
vectype, synth_shift_p))
return NULL;
tree accumulator;
/* Clear out the sequence of statements so we can populate it below. */
gimple *stmt = NULL;
if (cast_to_unsigned_p)
{
tree tmp_op = vect_recog_temp_ssa_var (multtype, NULL);
stmt = gimple_build_assign (tmp_op, CONVERT_EXPR, op);
append_pattern_def_seq (stmt_vinfo, stmt);
op = tmp_op;
}
if (alg.op[0] == alg_zero)
accumulator = build_int_cst (multtype, 0);
else
accumulator = op;
bool needs_fixup = (variant == negate_variant)
|| (variant == add_variant);
for (int i = 1; i < alg.ops; i++)
{
tree shft_log = build_int_cst (multtype, alg.log[i]);
tree accum_tmp = vect_recog_temp_ssa_var (multtype, NULL);
tree tmp_var = NULL_TREE;
switch (alg.op[i])
{
case alg_shift:
if (synth_shift_p)
stmt
= synth_lshift_by_additions (accum_tmp, accumulator, alg.log[i],
stmt_vinfo);
else
stmt = gimple_build_assign (accum_tmp, LSHIFT_EXPR, accumulator,
shft_log);
break;
case alg_add_t_m2:
tmp_var
= apply_binop_and_append_stmt (LSHIFT_EXPR, op, shft_log,
stmt_vinfo, synth_shift_p);
stmt = gimple_build_assign (accum_tmp, PLUS_EXPR, accumulator,
tmp_var);
break;
case alg_sub_t_m2:
tmp_var = apply_binop_and_append_stmt (LSHIFT_EXPR, op,
shft_log, stmt_vinfo,
synth_shift_p);
/* In some algorithms the first step involves zeroing the
accumulator. If subtracting from such an accumulator
just emit the negation directly. */
if (integer_zerop (accumulator))
stmt = gimple_build_assign (accum_tmp, NEGATE_EXPR, tmp_var);
else
stmt = gimple_build_assign (accum_tmp, MINUS_EXPR, accumulator,
tmp_var);
break;
case alg_add_t2_m:
tmp_var
= apply_binop_and_append_stmt (LSHIFT_EXPR, accumulator, shft_log,
stmt_vinfo, synth_shift_p);
stmt = gimple_build_assign (accum_tmp, PLUS_EXPR, tmp_var, op);
break;
case alg_sub_t2_m:
tmp_var
= apply_binop_and_append_stmt (LSHIFT_EXPR, accumulator, shft_log,
stmt_vinfo, synth_shift_p);
stmt = gimple_build_assign (accum_tmp, MINUS_EXPR, tmp_var, op);
break;
case alg_add_factor:
tmp_var
= apply_binop_and_append_stmt (LSHIFT_EXPR, accumulator, shft_log,
stmt_vinfo, synth_shift_p);
stmt = gimple_build_assign (accum_tmp, PLUS_EXPR, accumulator,
tmp_var);
break;
case alg_sub_factor:
tmp_var
= apply_binop_and_append_stmt (LSHIFT_EXPR, accumulator, shft_log,
stmt_vinfo, synth_shift_p);
stmt = gimple_build_assign (accum_tmp, MINUS_EXPR, tmp_var,
accumulator);
break;
default:
gcc_unreachable ();
}
/* We don't want to append the last stmt in the sequence to stmt_vinfo
but rather return it directly. */
if ((i < alg.ops - 1) || needs_fixup || cast_to_unsigned_p)
append_pattern_def_seq (stmt_vinfo, stmt);
accumulator = accum_tmp;
}
if (variant == negate_variant)
{
tree accum_tmp = vect_recog_temp_ssa_var (multtype, NULL);
stmt = gimple_build_assign (accum_tmp, NEGATE_EXPR, accumulator);
accumulator = accum_tmp;
if (cast_to_unsigned_p)
append_pattern_def_seq (stmt_vinfo, stmt);
}
else if (variant == add_variant)
{
tree accum_tmp = vect_recog_temp_ssa_var (multtype, NULL);
stmt = gimple_build_assign (accum_tmp, PLUS_EXPR, accumulator, op);
accumulator = accum_tmp;
if (cast_to_unsigned_p)
append_pattern_def_seq (stmt_vinfo, stmt);
}
/* Move back to a signed if needed. */
if (cast_to_unsigned_p)
{
tree accum_tmp = vect_recog_temp_ssa_var (itype, NULL);
stmt = gimple_build_assign (accum_tmp, CONVERT_EXPR, accumulator);
}
return stmt;
}
/* Detect multiplication by constant and convert it into a sequence of
shifts and additions, subtractions, negations. We reuse the
choose_mult_variant algorithms from expmed.c
Input/Output:
STMT_VINFO: The stmt from which the pattern search begins,
i.e. the mult stmt.
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace
the multiplication. */
static gimple *
vect_recog_mult_pattern (stmt_vec_info stmt_vinfo, tree *type_out)
{
vec_info *vinfo = stmt_vinfo->vinfo;
gimple *last_stmt = stmt_vinfo->stmt;
tree oprnd0, oprnd1, vectype, itype;
gimple *pattern_stmt;
if (!is_gimple_assign (last_stmt))
return NULL;
if (gimple_assign_rhs_code (last_stmt) != MULT_EXPR)
return NULL;
oprnd0 = gimple_assign_rhs1 (last_stmt);
oprnd1 = gimple_assign_rhs2 (last_stmt);
itype = TREE_TYPE (oprnd0);
if (TREE_CODE (oprnd0) != SSA_NAME
|| TREE_CODE (oprnd1) != INTEGER_CST
|| !INTEGRAL_TYPE_P (itype)
|| !type_has_mode_precision_p (itype))
return NULL;
vectype = get_vectype_for_scalar_type (vinfo, itype);
if (vectype == NULL_TREE)
return NULL;
/* If the target can handle vectorized multiplication natively,
don't attempt to optimize this. */
optab mul_optab = optab_for_tree_code (MULT_EXPR, vectype, optab_default);
if (mul_optab != unknown_optab)
{
machine_mode vec_mode = TYPE_MODE (vectype);
int icode = (int) optab_handler (mul_optab, vec_mode);
if (icode != CODE_FOR_nothing)
return NULL;
}
pattern_stmt = vect_synth_mult_by_constant (oprnd0, oprnd1, stmt_vinfo);
if (!pattern_stmt)
return NULL;
/* Pattern detected. */
vect_pattern_detected ("vect_recog_mult_pattern", last_stmt);
*type_out = vectype;
return pattern_stmt;
}
/* Detect a signed division by a constant that wouldn't be
otherwise vectorized:
type a_t, b_t;
S1 a_t = b_t / N;
where type 'type' is an integral type and N is a constant.
Similarly handle modulo by a constant:
S4 a_t = b_t % N;
Input/Output:
* STMT_VINFO: The stmt from which the pattern search begins,
i.e. the division stmt. S1 is replaced by if N is a power
of two constant and type is signed:
S3 y_t = b_t < 0 ? N - 1 : 0;
S2 x_t = b_t + y_t;
S1' a_t = x_t >> log2 (N);
S4 is replaced if N is a power of two constant and
type is signed by (where *_T temporaries have unsigned type):
S9 y_T = b_t < 0 ? -1U : 0U;
S8 z_T = y_T >> (sizeof (type_t) * CHAR_BIT - log2 (N));
S7 z_t = (type) z_T;
S6 w_t = b_t + z_t;
S5 x_t = w_t & (N - 1);
S4' a_t = x_t - z_t;
Output:
* TYPE_OUT: The type of the output of this pattern.
* Return value: A new stmt that will be used to replace the division
S1 or modulo S4 stmt. */
static gimple *
vect_recog_divmod_pattern (stmt_vec_info stmt_vinfo, tree *type_out)
{
vec_info *vinfo = stmt_vinfo->vinfo;
gimple *last_stmt = stmt_vinfo->stmt;
tree oprnd0, oprnd1, vectype, itype, cond;
gimple *pattern_stmt, *def_stmt;
enum tree_code rhs_code;
optab optab;
tree q;
int dummy_int, prec;
if (!is_gimple_assign (last_stmt))
return NULL;
rhs_code = gimple_assign_rhs_code (last_stmt);
switch (rhs_code)
{
case TRUNC_DIV_EXPR:
case EXACT_DIV_EXPR:
case TRUNC_MOD_EXPR:
break;
default:
return NULL;
}
oprnd0 = gimple_assign_rhs1 (last_stmt);
oprnd1 = gimple_assign_rhs2 (last_stmt);
itype = TREE_TYPE (oprnd0);
if (TREE_CODE (oprnd0) != SSA_NAME
|| TREE_CODE (oprnd1) != INTEGER_CST
|| TREE_CODE (itype) != INTEGER_TYPE
|| !type_has_mode_precision_p (itype))
return NULL;
scalar_int_mode itype_mode = SCALAR_INT_TYPE_MODE (itype);
vectype = get_vectype_for_scalar_type (vinfo, itype);
if (vectype == NULL_TREE)
return NULL;
if (optimize_bb_for_size_p (gimple_bb (last_stmt)))
{
/* If the target can handle vectorized division or modulo natively,
don't attempt to optimize this, since native division is likely
to give smaller code. */
optab = optab_for_tree_code (rhs_code, vectype, optab_default);
if (optab != unknown_optab)
{
machine_mode vec_mode = TYPE_MODE (vectype);
int icode = (int) optab_handler (optab, vec_mode);
if (icode != CODE_FOR_nothing)
return NULL;
}
}
prec = TYPE_PRECISION (itype);
if (integer_pow2p (oprnd1))
{
if (TYPE_UNSIGNED (itype) || tree_int_cst_sgn (oprnd1) != 1)
return NULL;
/* Pattern detected. */
vect_pattern_detected ("vect_recog_divmod_pattern", last_stmt);
*type_out = vectype;
/* Check if the target supports this internal function. */
internal_fn ifn = IFN_DIV_POW2;
if (direct_internal_fn_supported_p (ifn, vectype, OPTIMIZE_FOR_SPEED))
{
tree shift = build_int_cst (itype, tree_log2 (oprnd1));
tree var_div = vect_recog_temp_ssa_var (itype, NULL);
gimple *div_stmt = gimple_build_call_internal (ifn, 2, oprnd0, shift);
gimple_call_set_lhs (div_stmt, var_div);
if (rhs_code == TRUNC_MOD_EXPR)
{
append_pattern_def_seq (stmt_vinfo, div_stmt);
def_stmt
= gimple_build_assign (vect_recog_temp_ssa_var (itype, NULL),
LSHIFT_EXPR, var_div, shift);
append_pattern_def_seq (stmt_vinfo, def_stmt);
pattern_stmt
= gimple_build_assign (vect_recog_temp_ssa_var (itype, NULL),
MINUS_EXPR, oprnd0,
gimple_assign_lhs (def_stmt));
}
else
pattern_stmt = div_stmt;
gimple_set_location (pattern_stmt, gimple_location (last_stmt));
return pattern_stmt;
}
cond = build2 (LT_EXPR, boolean_type_node, oprnd0,
build_int_cst (itype, 0));
if (rhs_code == TRUNC_DIV_EXPR
|| rhs_code == EXACT_DIV_EXPR)
{
tree var = vect_recog_temp_ssa_var (itype, NULL);
tree shift;
def_stmt
= gimple_build_assign (var, COND_EXPR, cond,
fold_build2 (MINUS_EXPR, itype, oprnd1,
build_int_cst (itype, 1)),
build_int_cst (itype, 0));
append_pattern_def_seq (stmt_vinfo, def_stmt);
var = vect_recog_temp_ssa_var (itype, NULL);
def_stmt
= gimple_build_assign (var, PLUS_EXPR, oprnd0,
gimple_assign_lhs (def_stmt));
append_pattern_def_seq (stmt_vinfo, def_stmt);
shift = build_int_cst (itype, tree_log2 (oprnd1));
pattern_stmt
= gimple_build_assign (vect_recog_temp_ssa_var (itype, NULL),
RSHIFT_EXPR, var, shift);
}
else
{
tree signmask;
if (compare_tree_int (oprnd1, 2) == 0)
{
signmask = vect_recog_temp_ssa_var (itype, NULL);
def_stmt = gimple_build_assign (signmask, COND_EXPR, cond,
build_int_cst (itype, 1),
build_int_cst (itype, 0));
append_pattern_def_seq (stmt_vinfo, def_stmt);
}
else
{
tree utype
= build_nonstandard_integer_type (prec, 1);
tree vecutype = get_vectype_for_scalar_type (vinfo, utype);
tree shift
= build_int_cst (utype, GET_MODE_BITSIZE (itype_mode)
- tree_log2 (oprnd1));
tree var = vect_recog_temp_ssa_var (utype, NULL);
def_stmt = gimple_build_assign (var, COND_EXPR, cond,
build_int_cst (utype, -1),
build_int_cst (utype, 0));
append_pattern_def_seq (stmt_vinfo, def_stmt, vecutype);
var = vect_recog_temp_ssa_var (utype, NULL);
def_stmt = gimple_build_assign (var, RSHIFT_EXPR,
gimple_assign_lhs (def_stmt),
shift);
append_pattern_def_seq (stmt_vinfo, def_stmt, vecutype);
signmask = vect_recog_temp_ssa_var (itype, NULL);
def_stmt
= gimple_build_assign (signmask, NOP_EXPR, var);
append_pattern_def_seq (stmt_vinfo, def_stmt);
}
def_stmt
= gimple_build_assign (vect_recog_temp_ssa_var (itype, NULL),
PLUS_EXPR, oprnd0, signmask);
append_pattern_def_seq (stmt_vinfo, def_stmt);
def_stmt
= gimple_build_assign (vect_recog_temp_ssa_var (itype, NULL),
BIT_AND_EXPR, gimple_assign_lhs (def_stmt),
fold_build2 (MINUS_EXPR, itype, oprnd1,
build_int_cst (itype, 1)));
append_pattern_def_seq (stmt_vinfo, def_stmt);
pattern_stmt
= gimple_build_assign (vect_recog_temp_ssa_var (itype, NULL),
MINUS_EXPR, gimple_assign_lhs (def_stmt),
signmask);
}
return pattern_stmt;
}
if (prec > HOST_BITS_PER_WIDE_INT
|| integer_zerop (oprnd1))
return NULL;
if (!can_mult_highpart_p (TYPE_MODE (vectype), TYPE_UNSIGNED (itype)))
return NULL;
if (TYPE_UNSIGNED (itype))
{
unsigned HOST_WIDE_INT mh, ml;
int pre_shift, post_shift;
unsigned HOST_WIDE_INT d = (TREE_INT_CST_LOW (oprnd1)
& GET_MODE_MASK (itype_mode));
tree t1, t2, t3, t4;
if (d >= (HOST_WIDE_INT_1U << (prec - 1)))
/* FIXME: Can transform this into oprnd0 >= oprnd1 ? 1 : 0. */
return NULL;
/* Find a suitable multiplier and right shift count
instead of multiplying with D. */
mh = choose_multiplier (d, prec, prec, &ml, &post_shift, &dummy_int);
/* If the suggested multiplier is more than SIZE bits, we can do better
for even divisors, using an initial right shift. */
if (mh != 0 && (d & 1) == 0)
{
pre_shift = ctz_or_zero (d);
mh = choose_multiplier (d >> pre_shift, prec, prec - pre_shift,
&ml, &post_shift, &dummy_int);
gcc_assert (!mh);
}
else
pre_shift = 0;
if (mh != 0)
{
if (post_shift - 1 >= prec)
return NULL;
/* t1 = oprnd0 h* ml;
t2 = oprnd0 - t1;