Re: [PATCH] tree-optimization/115254 - don't account single-lane SLP against discovery limit

Re: [PATCH] tree-optimization/115254 - don't account single-lane SLP against discovery limit

[YunQiang Su]

Richard Biener <rguenther@suse.de> 于2024年5月28日周二 17:47写道：
>
> The following avoids accounting single-lane SLP to the discovery
> limit.  As the two testcases show this makes discovery fail,
> unfortunately even not the same across targets.  The following
> should fix two FAILs for GCN as a side-effect.
>
> Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
>
>         PR tree-optimization/115254
>         * tree-vect-slp.cc (vect_build_slp_tree): Only account
>         multi-lane SLP to limit.
>
>         * gcc.dg/vect/slp-cond-2-big-array.c: Expect 4 times SLP.
>         * gcc.dg/vect/slp-cond-2.c: Likewise.

With this patch, MIPS/MSA still has only 3 times SLP.
I am digging the problem


> ---
>  .../gcc.dg/vect/slp-cond-2-big-array.c        |  2 +-
>  gcc/testsuite/gcc.dg/vect/slp-cond-2.c        |  2 +-
>  gcc/tree-vect-slp.cc                          | 31 +++++++++++--------
>  3 files changed, 20 insertions(+), 15 deletions(-)
>
> diff --git a/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c b/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
> index cb7eb94b3a3..9a9f63c0b8d 100644
> --- a/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
> +++ b/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
> @@ -128,4 +128,4 @@ main ()
>    return 0;
>  }
>
> -/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 3 "vect" } } */
> +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect" } } */
> diff --git a/gcc/testsuite/gcc.dg/vect/slp-cond-2.c b/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
> index 1dcee46cd95..08bbb3dbec6 100644
> --- a/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
> +++ b/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
> @@ -128,4 +128,4 @@ main ()
>    return 0;
>  }
>
> -/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 3 "vect" } } */
> +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect" } } */
> diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
> index 0dd9a4daf6a..bbfde8849c1 100644
> --- a/gcc/tree-vect-slp.cc
> +++ b/gcc/tree-vect-slp.cc
> @@ -1725,21 +1725,26 @@ vect_build_slp_tree (vec_info *vinfo,
>    SLP_TREE_SCALAR_STMTS (res) = stmts;
>    bst_map->put (stmts.copy (), res);
>
> -  if (*limit == 0)
> +  /* Single-lane SLP doesn't have the chance of run-away, do not account
> +     it to the limit.  */
> +  if (stmts.length () > 1)
>      {
> -      if (dump_enabled_p ())
> -       dump_printf_loc (MSG_NOTE, vect_location,
> -                        "SLP discovery limit exceeded\n");
> -      /* Mark the node invalid so we can detect those when still in use
> -        as backedge destinations.  */
> -      SLP_TREE_SCALAR_STMTS (res) = vNULL;
> -      SLP_TREE_DEF_TYPE (res) = vect_uninitialized_def;
> -      res->failed = XNEWVEC (bool, group_size);
> -      memset (res->failed, 0, sizeof (bool) * group_size);
> -      memset (matches, 0, sizeof (bool) * group_size);
> -      return NULL;
> +      if (*limit == 0)
> +       {
> +         if (dump_enabled_p ())
> +           dump_printf_loc (MSG_NOTE, vect_location,
> +                            "SLP discovery limit exceeded\n");
> +         /* Mark the node invalid so we can detect those when still in use
> +            as backedge destinations.  */
> +         SLP_TREE_SCALAR_STMTS (res) = vNULL;
> +         SLP_TREE_DEF_TYPE (res) = vect_uninitialized_def;
> +         res->failed = XNEWVEC (bool, group_size);
> +         memset (res->failed, 0, sizeof (bool) * group_size);
> +         memset (matches, 0, sizeof (bool) * group_size);
> +         return NULL;
> +       }
> +      --*limit;
>      }
> -  --*limit;
>
>    if (dump_enabled_p ())
>      dump_printf_loc (MSG_NOTE, vect_location,
> --
> 2.35.3



-- 
YunQiang Su

Re: [PATCH] tree-optimization/115254 - don't account single-lane SLP against discovery limit

[Richard Biener]

[-- Attachment #1: Type: text/plain, Size: 4264 bytes --]

On Thu, 6 Jun 2024, YunQiang Su wrote:

> Richard Biener <rguenther@suse.de> 于2024年5月28日周二 17:47写道：
> >
> > The following avoids accounting single-lane SLP to the discovery
> > limit.  As the two testcases show this makes discovery fail,
> > unfortunately even not the same across targets.  The following
> > should fix two FAILs for GCN as a side-effect.
> >
> > Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
> >
> >         PR tree-optimization/115254
> >         * tree-vect-slp.cc (vect_build_slp_tree): Only account
> >         multi-lane SLP to limit.
> >
> >         * gcc.dg/vect/slp-cond-2-big-array.c: Expect 4 times SLP.
> >         * gcc.dg/vect/slp-cond-2.c: Likewise.
> 
> With this patch, MIPS/MSA still has only 3 times SLP.
> I am digging the problem

I bet it's an issue with missed permutes.  f3() requires interleaving
of two VnQImode vectors.

> 
> > ---
> >  .../gcc.dg/vect/slp-cond-2-big-array.c        |  2 +-
> >  gcc/testsuite/gcc.dg/vect/slp-cond-2.c        |  2 +-
> >  gcc/tree-vect-slp.cc                          | 31 +++++++++++--------
> >  3 files changed, 20 insertions(+), 15 deletions(-)
> >
> > diff --git a/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c b/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
> > index cb7eb94b3a3..9a9f63c0b8d 100644
> > --- a/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
> > +++ b/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
> > @@ -128,4 +128,4 @@ main ()
> >    return 0;
> >  }
> >
> > -/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 3 "vect" } } */
> > +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect" } } */
> > diff --git a/gcc/testsuite/gcc.dg/vect/slp-cond-2.c b/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
> > index 1dcee46cd95..08bbb3dbec6 100644
> > --- a/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
> > +++ b/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
> > @@ -128,4 +128,4 @@ main ()
> >    return 0;
> >  }
> >
> > -/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 3 "vect" } } */
> > +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect" } } */
> > diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
> > index 0dd9a4daf6a..bbfde8849c1 100644
> > --- a/gcc/tree-vect-slp.cc
> > +++ b/gcc/tree-vect-slp.cc
> > @@ -1725,21 +1725,26 @@ vect_build_slp_tree (vec_info *vinfo,
> >    SLP_TREE_SCALAR_STMTS (res) = stmts;
> >    bst_map->put (stmts.copy (), res);
> >
> > -  if (*limit == 0)
> > +  /* Single-lane SLP doesn't have the chance of run-away, do not account
> > +     it to the limit.  */
> > +  if (stmts.length () > 1)
> >      {
> > -      if (dump_enabled_p ())
> > -       dump_printf_loc (MSG_NOTE, vect_location,
> > -                        "SLP discovery limit exceeded\n");
> > -      /* Mark the node invalid so we can detect those when still in use
> > -        as backedge destinations.  */
> > -      SLP_TREE_SCALAR_STMTS (res) = vNULL;
> > -      SLP_TREE_DEF_TYPE (res) = vect_uninitialized_def;
> > -      res->failed = XNEWVEC (bool, group_size);
> > -      memset (res->failed, 0, sizeof (bool) * group_size);
> > -      memset (matches, 0, sizeof (bool) * group_size);
> > -      return NULL;
> > +      if (*limit == 0)
> > +       {
> > +         if (dump_enabled_p ())
> > +           dump_printf_loc (MSG_NOTE, vect_location,
> > +                            "SLP discovery limit exceeded\n");
> > +         /* Mark the node invalid so we can detect those when still in use
> > +            as backedge destinations.  */
> > +         SLP_TREE_SCALAR_STMTS (res) = vNULL;
> > +         SLP_TREE_DEF_TYPE (res) = vect_uninitialized_def;
> > +         res->failed = XNEWVEC (bool, group_size);
> > +         memset (res->failed, 0, sizeof (bool) * group_size);
> > +         memset (matches, 0, sizeof (bool) * group_size);
> > +         return NULL;
> > +       }
> > +      --*limit;
> >      }
> > -  --*limit;
> >
> >    if (dump_enabled_p ())
> >      dump_printf_loc (MSG_NOTE, vect_location,
> > --
> > 2.35.3
> 
> 
> 
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)

Re: [PATCH] tree-optimization/115254 - don't account single-lane SLP against discovery limit

[YunQiang Su]

Richard Biener <rguenther@suse.de> 于2024年6月6日周四 14:20写道：
>
> On Thu, 6 Jun 2024, YunQiang Su wrote:
>
> > Richard Biener <rguenther@suse.de> 于2024年5月28日周二 17:47写道：
> > >
> > > The following avoids accounting single-lane SLP to the discovery
> > > limit.  As the two testcases show this makes discovery fail,
> > > unfortunately even not the same across targets.  The following
> > > should fix two FAILs for GCN as a side-effect.
> > >
> > > Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
> > >
> > >         PR tree-optimization/115254
> > >         * tree-vect-slp.cc (vect_build_slp_tree): Only account
> > >         multi-lane SLP to limit.
> > >
> > >         * gcc.dg/vect/slp-cond-2-big-array.c: Expect 4 times SLP.
> > >         * gcc.dg/vect/slp-cond-2.c: Likewise.
> >
> > With this patch, MIPS/MSA still has only 3 times SLP.
> > I am digging the problem
>
> I bet it's an issue with missed permutes.  f3() requires interleaving
> of two VnQImode vectors.
>

Thanks. This problem disappears when I try to implement vcond_mask.

[PATCH] tree-optimization/115254 - don't account single-lane SLP against discovery limit

[Richard Biener]

The following avoids accounting single-lane SLP to the discovery
limit.  As the two testcases show this makes discovery fail,
unfortunately even not the same across targets.  The following
should fix two FAILs for GCN as a side-effect.

Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.

	PR tree-optimization/115254
	* tree-vect-slp.cc (vect_build_slp_tree): Only account
	multi-lane SLP to limit.

	* gcc.dg/vect/slp-cond-2-big-array.c: Expect 4 times SLP.
	* gcc.dg/vect/slp-cond-2.c: Likewise.
---
 .../gcc.dg/vect/slp-cond-2-big-array.c        |  2 +-
 gcc/testsuite/gcc.dg/vect/slp-cond-2.c        |  2 +-
 gcc/tree-vect-slp.cc                          | 31 +++++++++++--------
 3 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c b/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
index cb7eb94b3a3..9a9f63c0b8d 100644
--- a/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
+++ b/gcc/testsuite/gcc.dg/vect/slp-cond-2-big-array.c
@@ -128,4 +128,4 @@ main ()
   return 0;
 }
 
-/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 3 "vect" } } */
+/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect" } } */
diff --git a/gcc/testsuite/gcc.dg/vect/slp-cond-2.c b/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
index 1dcee46cd95..08bbb3dbec6 100644
--- a/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
+++ b/gcc/testsuite/gcc.dg/vect/slp-cond-2.c
@@ -128,4 +128,4 @@ main ()
   return 0;
 }
 
-/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 3 "vect" } } */
+/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect" } } */
diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
index 0dd9a4daf6a..bbfde8849c1 100644
--- a/gcc/tree-vect-slp.cc
+++ b/gcc/tree-vect-slp.cc
@@ -1725,21 +1725,26 @@ vect_build_slp_tree (vec_info *vinfo,
   SLP_TREE_SCALAR_STMTS (res) = stmts;
   bst_map->put (stmts.copy (), res);
 
-  if (*limit == 0)
+  /* Single-lane SLP doesn't have the chance of run-away, do not account
+     it to the limit.  */
+  if (stmts.length () > 1)
     {
-      if (dump_enabled_p ())
-	dump_printf_loc (MSG_NOTE, vect_location,
-			 "SLP discovery limit exceeded\n");
-      /* Mark the node invalid so we can detect those when still in use
-	 as backedge destinations.  */
-      SLP_TREE_SCALAR_STMTS (res) = vNULL;
-      SLP_TREE_DEF_TYPE (res) = vect_uninitialized_def;
-      res->failed = XNEWVEC (bool, group_size);
-      memset (res->failed, 0, sizeof (bool) * group_size);
-      memset (matches, 0, sizeof (bool) * group_size);
-      return NULL;
+      if (*limit == 0)
+	{
+	  if (dump_enabled_p ())
+	    dump_printf_loc (MSG_NOTE, vect_location,
+			     "SLP discovery limit exceeded\n");
+	  /* Mark the node invalid so we can detect those when still in use
+	     as backedge destinations.  */
+	  SLP_TREE_SCALAR_STMTS (res) = vNULL;
+	  SLP_TREE_DEF_TYPE (res) = vect_uninitialized_def;
+	  res->failed = XNEWVEC (bool, group_size);
+	  memset (res->failed, 0, sizeof (bool) * group_size);
+	  memset (matches, 0, sizeof (bool) * group_size);
+	  return NULL;
+	}
+      --*limit;
     }
-  --*limit;
 
   if (dump_enabled_p ())
     dump_printf_loc (MSG_NOTE, vect_location,
-- 
2.35.3