Richard Biener <rguent...@suse.de> writes:
> This adds a --param to allow disabling of vectorization of
> floating point inductions. Ontop of -Ofast this should allow
> 549.fotonik3d_r to not miscompare.
>
> While I thought of a more elaborate way of disabling certain
> vectorization kinds (reductions also came to my mind) this
> for now simply uses a --param than some sophisticated -fvectorize-*
> scheme.
>
> Bootstrapped and tested on x86_64-unknown-linux-gnu. I've
> verified that 549.fotonik3d_r miscompares with -Ofast -march=znver2
> and passes when adding --param vect-induction-float=0 which
> should be valid at least for peak (but I guess also base for
> FOPTIMIZE for example). I did not benchmark against other
> workarounds (it has been said -fno-unsafe-math-optimizations
> or other similar things work as well).
>
> OK for trunk?
>
> Thanks,
> Richard.
>
> 2022-03-08 Richard Biener <rguent...@suse.de>
>
> PR tree-optimization/84201
> * params.opt (-param=vect-induction-float): Add.
> * doc/invoke.texi (vect-induction-float): Document.
> * tree-vect-loop.cc (vectorizable_induction): Honor
> param_vect_induction_float.
>
> * gcc.dg/vect/pr84201.c: New testcase.
LGTM FWIW.
Thanks,
Richard
> ---
> gcc/doc/invoke.texi | 3 +++
> gcc/params.opt | 4 ++++
> gcc/testsuite/gcc.dg/vect/pr84201.c | 22 ++++++++++++++++++++++
> gcc/tree-vect-loop.cc | 8 ++++++++
> 4 files changed, 37 insertions(+)
> create mode 100644 gcc/testsuite/gcc.dg/vect/pr84201.c
>
> diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
> index b01ffab566a..a0fa5e1cf43 100644
> --- a/gcc/doc/invoke.texi
> +++ b/gcc/doc/invoke.texi
> @@ -14989,6 +14989,9 @@ in an inner loop relative to the loop being
> vectorized. The factor applied
> is the maximum of the estimated number of iterations of the inner loop and
> this parameter. The default value of this parameter is 50.
>
> +@item vect-induction-float
> +Enable loop vectorization of floating point inductions.
> +
> @item avoid-fma-max-bits
> Maximum number of bits for which we avoid creating FMAs.
>
> diff --git a/gcc/params.opt b/gcc/params.opt
> index f76f7839916..9561aa61a50 100644
> --- a/gcc/params.opt
> +++ b/gcc/params.opt
> @@ -1176,6 +1176,10 @@ Controls how loop vectorizer uses partial vectors. 0
> means never, 1 means only
> Common Joined UInteger Var(param_vect_inner_loop_cost_factor) Init(50)
> IntegerRange(1, 10000) Param Optimization
> The maximum factor which the loop vectorizer applies to the cost of
> statements in an inner loop relative to the loop being vectorized.
>
> +-param=vect-induction-float=
> +Common Joined UInteger Var(param_vect_induction_float) Init(1)
> IntegerRage(0, 1) Param Optimization
> +Enable loop vectorization of floating point inductions.
> +
> -param=vrp1-mode=
> Common Joined Var(param_vrp1_mode) Enum(vrp_mode) Init(VRP_MODE_VRP) Param
> Optimization
> --param=vrp1-mode=[vrp|ranger] Specifies the mode VRP1 should operate in.
> diff --git a/gcc/testsuite/gcc.dg/vect/pr84201.c
> b/gcc/testsuite/gcc.dg/vect/pr84201.c
> new file mode 100644
> index 00000000000..1cc6d1ff13c
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/vect/pr84201.c
> @@ -0,0 +1,22 @@
> +/* { dg-do compile } */
> +/* { dg-additional-options "-Ofast --param vect-induction-float=0" } */
> +
> +void foo (float *a, float f, float s, int n)
> +{
> + for (int i = 0; i < n; ++i)
> + {
> + a[i] = f;
> + f += s;
> + }
> +}
> +
> +void bar (double *a, double f, double s, int n)
> +{
> + for (int i = 0; i < n; ++i)
> + {
> + a[i] = f;
> + f += s;
> + }
> +}
> +
> +/* { dg-final { scan-tree-dump-times "vectorized 0 loops" 2 "vect" } } */
> diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
> index 1f30fc82ca1..7fcec12a3e9 100644
> --- a/gcc/tree-vect-loop.cc
> +++ b/gcc/tree-vect-loop.cc
> @@ -8175,6 +8175,14 @@ vectorizable_induction (loop_vec_info loop_vinfo,
> return false;
> }
>
> + if (FLOAT_TYPE_P (vectype) && !param_vect_induction_float)
> + {
> + if (dump_enabled_p ())
> + dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
> + "floating point induction vectorization disabled\n");
> + return false;
> + }
> +
> step_expr = STMT_VINFO_LOOP_PHI_EVOLUTION_PART (stmt_info);
> gcc_assert (step_expr != NULL_TREE);
> tree step_vectype = get_same_sized_vectype (TREE_TYPE (step_expr),
> vectype);
