> -----Original Message-----
> From: Richard Biener <[email protected]>
> Sent: 05 October 2026 12:23
> To: Tamar Christina <[email protected]>
> Cc: [email protected]; nd <[email protected]>; [email protected]
> Subject: Re: [patch][vect]: don't create unsupported reduction recurrence in
> ifcvt [PR126551]
>
> On Mon, 5 Oct 2026, Tamar Christina wrote:
>
> > The example
> >
> > float
> > f (const float *x, int n)
> > {
> > float acc = 0;
> > for (int i = 0; i < n; ++i)
> > {
> > float d = x[i];
> > if (d < 0)
> > acc += d;
> > acc += d * 3.0f;
> > }
> > return acc;
> > }
> >
> > compiled at -Ofast fails to vectorize with
> >
> > note: vect_is_simple_use: operand acc_17 = PHI <acc_13(8), 0.0(18)>,
> > type of def: unknown
> > missed: Unsupported pattern.
> > missed: not vectorized: unsupported use in stmt.
> >
> > due to
> >
> > note: Analyze phi: acc_17 = PHI <acc_13(8), 0.0(18)>
> > missed: reduction used in loop.
> > missed: Unknown def-use cycle pattern.
> >
> > This is happening because before if-convert we have the following reduction:
> >
> > acc_1 = PHI <acc_0, acc_2>
> > if (cond)
> > acc_2 = acc_1 + value;
> >
> > and so the reduction value is updated only on some paths.
> >
> > if-cvt changes this into
> >
> > acc_3 = cond ? acc_2 : acc_1;
> >
> > but since acc_2 itself is acc_1 + value this is actually
> >
> > acc_3 = cond ? acc_1 + value : acc_1;
> >
> > This is semantically sound but hides the recurrence from the vectorizer.
> >
> > In principle if-cvt creates
> >
> > acc_1 = PHI <0, acc_3>
> > acc_2 = acc_1 + value
> > acc_3 = cond ? acc_2 : acc_1
> >
> > so the loop carried value acc_3 is inside the COND_EXPR and acc_1 is used
> twice
> > inside the same reduction. The update chain is
> >
> > PHI -> PLUS -> COND_EXPR -> PHI
> > \--------------^
> >
> > which is a problem since reduction analysis needs a single statement that
> > updates the loop carried accumulator.
> >
> > Note that if the user supplied this form manually, if-cvt rewrites it to
> > one of
> > two forms that the vectorizer can handle. It either rewrites it to COND_ADD
> and
> > this removing the PLUS + COND_EXPR or it floats out the cond_expr.
> >
> > This patch does the latter, as picking COND_ADD here is premature, it forces
> > selection of a masked target even though the vectorizer these days can
> convert
> > the cond expr to a masked operation on its own based on costing.
> >
> > Essentially we change
> >
> > cond ? base + addend : base or cond ? base : base + addend
> >
> > into
> >
> > base + (cond ? addend : 0) or base + (cond ? 0 : addend)
> >
> > this turns the recurrence into a form the vectorizer can handle.
>
> This seems to be a candidate for a (match ...) and I wonder if it
> wouldn't be profitable even for non-reductions which should
> simplify all the checks?
>
Yeah, using a match if fair. I was specifically avoiding it to reduce the
number of reviews needed :) But will use it.
As for the non-reductions, I did have this before, but that in some
cases that created a form that when vectorized made multiplication
widening miss creating COND_FMAs.
So e.g. gcc.target/aarch64/sve/cond_mla_3.c and others had failures.
I didn't look into why, but can if you rather apply it broadly.
> Also the validity condition wouldn't be -fassociative-math
> but only that we can perform an addition of zero? (we could
> make that correct by adding -0.0 instead I think but then it
> might no longer be always profitable since -0.0 is more expensive
> to generate).
True, I added it because predicate_scalar_phi can swap operands
when the condition condition is a negated condition to simplify the
cond. And so I figured if I move the cond condition I might have swapped
the addition order.
I thought about -0.0, and was thinking that maybe I should gate it by
HONOR_SIGNED_ZEROS. So maybe when ! HONOR_SIGNED_ZEROS use
0.0 and if not -0.0, since without the transform the vectorizer is going to
bail out anyway.
Would that work for you?
Thanks,
Tamar
>
> > Special care is taken to only do this when the reduction is within the same
> > loop and where there is no other uses of the PHI reduction but the PHI node.
> >
> > i.e. the rewrite is only applied when BASE is the loop-carried reduction PHI
> for
> > the current loop and the if-converted PHI feeds the latch update for that
> same
> > reduction. This avoids changing unrelated conditional expressions or cases
> > where the apparent recurrence is not the loop reduction being analyzed.
> >
> > Bootstrapped Regtested on aarch64-none-linux-gnu,
> > arm-none-linux-gnueabihf, x86_64-pc-linux-gnu -m32, -m64 and no issues.
> >
> > Any comments?
> >
> > Thanks,
> > Tamar
> >
> > gcc/ChangeLog:
> >
> > PR tree-optimization/126551
> > * tree-if-conv.cc (ifc_temp_var): Support cases with no vUSE.
> > (ifc_plus_with_operand_p, ifc_phi_feeds_loop_reduction_p,
> > fold_build_cond_reduction_expr): New.
> > (predicate_scalar_phi): Use it.
> >
> > gcc/testsuite/ChangeLog:
> >
> > PR tree-optimization/126551
> > * gcc.dg/vect/vect-switch-ifcvt-2.c: Update testcases.
> > * gcc.dg/vect/cond-reduc-ifcvt-1.c: New test.
> > * gcc.dg/vect/cond-reduc-ifcvt-2.c: New test.
> >
> > ---
> > diff --git a/gcc/testsuite/gcc.dg/vect/cond-reduc-ifcvt-1.c
> b/gcc/testsuite/gcc.dg/vect/cond-reduc-ifcvt-1.c
> > new file mode 100644
> > index
> 0000000000000000000000000000000000000000..e4a2e7634a8d6bf02d
> 1d3a6f7f12515e83fb8423
> > --- /dev/null
> > +++ b/gcc/testsuite/gcc.dg/vect/cond-reduc-ifcvt-1.c
> > @@ -0,0 +1,21 @@
> > +/* { dg-do compile } */
> > +/* { dg-require-effective-target vect_float } */
> > +/* { dg-additional-options "-Ofast -fdump-tree-ifcvt-details" } */
> > +
> > +float
> > +f (const float *x, int n)
> > +{
> > + float acc = 0;
> > + for (int i = 0; i < n; ++i)
> > + {
> > + float d = x[i];
> > + if (d < 0)
> > + acc += d;
> > + acc += d * 3.0f;
> > + }
> > + return acc;
> > +}
> > +
> > +/* { dg-final { scan-tree-dump "new phi replacement stmt" "ifcvt" } } */
> > +/* { dg-final { scan-tree-dump "_ifc_.* = MIN_EXPR" "ifcvt" } } */
> > +/* { dg-final { scan-tree-dump " = .* \\+ _ifc_" "ifcvt" } } */
> > diff --git a/gcc/testsuite/gcc.dg/vect/cond-reduc-ifcvt-2.c
> b/gcc/testsuite/gcc.dg/vect/cond-reduc-ifcvt-2.c
> > new file mode 100644
> > index
> 0000000000000000000000000000000000000000..c6015303e5568dd51a
> 7ced1edf33784fdf93f75d
> > --- /dev/null
> > +++ b/gcc/testsuite/gcc.dg/vect/cond-reduc-ifcvt-2.c
> > @@ -0,0 +1,21 @@
> > +/* { dg-do compile } */
> > +/* { dg-require-effective-target vect_int } */
> > +/* { dg-additional-options "-O3 -fdump-tree-ifcvt-details" } */
> > +
> > +int
> > +f (const int *x, int n)
> > +{
> > + int acc = 0;
> > + for (int i = 0; i < n; ++i)
> > + {
> > + int d = x[i];
> > + if (d < 0)
> > + acc += d;
> > + acc += d * 3;
> > + }
> > + return acc;
> > +}
> > +
> > +/* { dg-final { scan-tree-dump "new phi replacement stmt" "ifcvt" } } */
> > +/* { dg-final { scan-tree-dump "_ifc_.* = MIN_EXPR" "ifcvt" } } */
> > +/* { dg-final { scan-tree-dump " = .* \\+ _ifc_" "ifcvt" } } */
> > diff --git a/gcc/testsuite/gcc.dg/vect/vect-switch-ifcvt-2.c
> b/gcc/testsuite/gcc.dg/vect/vect-switch-ifcvt-2.c
> > index
> f0fcb43e17702492b4fb0e9bc89f0c90b3106fda..03f241aef3750be06e91b7
> c9c583d4bb091e87e7 100644
> > --- a/gcc/testsuite/gcc.dg/vect/vect-switch-ifcvt-2.c
> > +++ b/gcc/testsuite/gcc.dg/vect/vect-switch-ifcvt-2.c
> > @@ -46,4 +46,4 @@ f2 (char *s)
> > return c;
> > }
> >
> > -/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 0 "vect" } } */
> > +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */
> > diff --git a/gcc/tree-if-conv.cc b/gcc/tree-if-conv.cc
> > index
> cc3e1dfb4ef2bbd1edaa5ce188202650f5cb163d..1fbea5383fa1914031fceb6
> 833ac7127055c8a81 100644
> > --- a/gcc/tree-if-conv.cc
> > +++ b/gcc/tree-if-conv.cc
> > @@ -392,7 +392,9 @@ ifc_temp_var (tree type, tree expr,
> gimple_stmt_iterator *gsi)
> > {
> > tree new_name = make_temp_ssa_name (type, NULL, "_ifc_");
> > gimple *stmt = gimple_build_assign (new_name, expr);
> > - gimple_set_vuse (stmt, gimple_vuse (gsi_stmt (*gsi)));
> > + gimple *curr_stmt = gsi_stmt (*gsi);
> > + if (curr_stmt)
> > + gimple_set_vuse (stmt, gimple_vuse (curr_stmt));
> > gsi_insert_before (gsi, stmt, GSI_SAME_STMT);
> > return new_name;
> > }
> > @@ -526,6 +528,117 @@ fold_build_cond_expr (tree type, tree cond, tree
> rhs, tree lhs)
> > return build3 (COND_EXPR, type, cond, rhs, lhs);
> > }
> >
> > +/* If EXPR is BASE + ADDEND, return true and store ADDEND. */
> > +
> > +static bool
> > +ifc_plus_with_operand_p (tree expr, tree base, tree *addend)
> > +{
> > + if (TREE_CODE (expr) != SSA_NAME)
> > + return false;
> > +
> > + gimple_match_op op;
> > + if (!gimple_extract_op (SSA_NAME_DEF_STMT (expr), &op)
> > + || op.code != PLUS_EXPR
> > + || op.num_ops != 2)
> > + return false;
> > +
> > + if (op.ops[0] == base)
> > + {
> > + *addend = op.ops[1];
> > + return true;
> > + }
> > + if (op.ops[1] == base)
> > + {
> > + *addend = op.ops[0];
> > + return true;
> > + }
> > + return false;
> > +}
> > +
> > +/* Return true if BASE is a loop reduction PHI and PHI feeds the latch
> > update
> > + for that reduction. */
> > +
> > +static bool
> > +ifc_phi_feeds_loop_reduction_p (gphi *phi, tree base)
> > +{
> > + use_operand_p use_p;
> > + gimple *use_stmt;
> > +
> > + if (TREE_CODE (base) != SSA_NAME
> > + || gimple_code (SSA_NAME_DEF_STMT (base)) != GIMPLE_PHI
> > + || !single_imm_use (gimple_phi_result (phi), &use_p, &use_stmt)
> > + || !is_gimple_assign (use_stmt))
> > + return false;
> > +
> > + gphi *header_phi = as_a <gphi *> (SSA_NAME_DEF_STMT (base));
> > + class loop *loop = gimple_bb (phi)->loop_father;
> > + if (gimple_bb (header_phi) != loop->header)
> > + return false;
> > +
> > + tree use_lhs = gimple_assign_lhs (use_stmt);
> > + edge latch_e = loop_latch_edge (loop);
> > + return PHI_ARG_DEF_FROM_EDGE (header_phi, latch_e) == use_lhs;
> > +}
> > +
> > +/* Return a vectorizer-friendly form of COND ? RHS : LHS if it selects
> > + between BASE + ADDEND and BASE. This keeps the reduction value in a
> > + single add chain after if-conversion:
> > +
> > + COND ? BASE + ADDEND : BASE -> BASE + (COND ? ADDEND : 0)
> > + COND ? BASE : BASE + ADDEND -> BASE + (COND ? 0 : ADDEND)
> > +
> > + PHI is the if-converted PHI whose result is being replaced. GSI is the
> > + insertion point for the temporary conditional addend.
> > +
> > + For floating-point types this reassociation is only done when
> > + -fassociative-math allows it. For integer types it does not introduce
> > new
> > + overflow: the selected ADDEND is either the original addend or zero, so
> the
> > + true arm keeps the original BASE + ADDEND and the false arm evaluates
> > + BASE + 0. Return null if the expression cannot be rewritten safely. */
> > +
> > +static tree
> > +fold_build_cond_reduction_expr (gphi *phi, tree type, tree cond, tree rhs,
> > + tree lhs, gimple_stmt_iterator *gsi)
> > +{
> > + tree addend;
> > + if (SCALAR_FLOAT_TYPE_P (type))
> > + {
> > + if (!flag_associative_math)
> > + return NULL_TREE;
> > + }
> > + else if (ANY_INTEGRAL_TYPE_P (type))
> > + {
> > + if (TYPE_OVERFLOW_TRAPS (type))
> > + return NULL_TREE;
> > + }
> > + else
> > + return NULL_TREE;
> > +
> > + if (ifc_plus_with_operand_p (rhs, lhs, &addend)
> > + && ifc_phi_feeds_loop_reduction_p (phi, lhs)
> > + && types_compatible_p (type, TREE_TYPE (lhs))
> > + && types_compatible_p (type, TREE_TYPE (addend)))
> > + {
> > + tree zero = build_zero_cst (type);
> > + tree expr = fold_build_cond_expr (type, cond, addend, zero);
> > + tree tmp = ifc_temp_var (type, expr, gsi);
> > + return build2 (PLUS_EXPR, type, lhs, tmp);
> > + }
> > +
> > + if (ifc_plus_with_operand_p (lhs, rhs, &addend)
> > + && ifc_phi_feeds_loop_reduction_p (phi, rhs)
> > + && types_compatible_p (type, TREE_TYPE (rhs))
> > + && types_compatible_p (type, TREE_TYPE (addend)))
> > + {
> > + tree zero = build_zero_cst (type);
> > + tree expr = fold_build_cond_expr (type, cond, zero, addend);
> > + tree tmp = ifc_temp_var (type, expr, gsi);
> > + return build2 (PLUS_EXPR, type, rhs, tmp);
> > + }
> > +
> > + return NULL_TREE;
> > +}
> > +
> > /* Add condition NC to the predicate list of basic block BB. LOOP is
> > the loop to be if-converted. Use predicate of cd-equivalent block
> > for join bb if it exists: we call basic blocks bb1 and bb2
> > @@ -2488,8 +2601,14 @@ predicate_scalar_phi (gphi *phi,
> gimple_stmt_iterator *gsi)
> > }
> > else
> > /* Build new RHS using selected condition and arguments. */
> > - rhs = fold_build_cond_expr (TREE_TYPE (res), unshare_expr (cond),
> > - arg0, arg1);
> > + {
> > + tree type = TREE_TYPE (res);
> > + rhs = fold_build_cond_reduction_expr (phi, type, cond, arg0,
> > + arg1, gsi);
> > + if (!rhs)
> > + rhs = fold_build_cond_expr (type, unshare_expr (cond), arg0,
> > + arg1);
> > + }
> > new_stmt = gimple_build_assign (res, rhs);
> > gsi_insert_before (gsi, new_stmt, GSI_SAME_STMT);
> > gimple_stmt_iterator new_gsi = gsi_for_stmt (new_stmt);
> > @@ -2574,9 +2693,16 @@ predicate_scalar_phi (gphi *phi,
> gimple_stmt_iterator *gsi)
> > NULL_TREE, true, GSI_SAME_STMT);
> > if (!(is_cond_scalar_reduction (phi, &reduc, arg0 , arg1,
> > &op0, &op1, true, &has_nop,
> &nop_reduc)))
> > - rhs = fold_build_cond_expr (TREE_TYPE (res), unshare_expr (cond),
> > - swap ? arg1 : arg0,
> > - swap ? arg0 : arg1);
> > + {
> > + tree type = TREE_TYPE (res);
> > + tree rhs_arg = swap ? arg1 : arg0;
> > + tree lhs_arg = swap ? arg0 : arg1;
> > + rhs = fold_build_cond_reduction_expr (phi, type, cond, rhs_arg,
> > + lhs_arg, gsi);
> > + if (!rhs)
> > + rhs = fold_build_cond_expr (type, unshare_expr (cond), rhs_arg,
> > + lhs_arg);
> > + }
> > else
> > {
> > /* Convert reduction stmt into vectorizable form. */
> >
> >
> >
>
> --
> Richard Biener <[email protected]>
> SUSE Software Solutions Germany GmbH,
> Frankenstrasse 146, 90461 Nuernberg, Germany;
> GF: Stefan Gaiser, Jochen Jaser, Abhinav Puri; (HRB 36809, AG Nuernberg)
