public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r11-10480] tree-optimization/107212 - SLP reduction of reduction paths
@ 2023-01-24 15:22 Richard Biener
  0 siblings, 0 replies; only message in thread
From: Richard Biener @ 2023-01-24 15:22 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:2461fa40fc24a403dc8149338f44b0e4aef4f173

commit r11-10480-g2461fa40fc24a403dc8149338f44b0e4aef4f173
Author: Richard Biener <rguenther@suse.de>
Date:   Tue Oct 11 11:34:55 2022 +0200

    tree-optimization/107212 - SLP reduction of reduction paths
    
    The following fixes an issue with how we handle epilogue generation
    for SLP reductions of reduction paths where the actual live lanes
    are not "canonical".  We need to make sure to identify all live
    lanes as reductions and thus have to iterate over all participating
    SLP lanes when walking the reduction SSA use-def chain.  Also the
    previous attempt likely to mitigate such issue in
    vectorizable_live_operation is misguided and has to be removed.
    
            PR tree-optimization/107212
            * tree-vect-loop.c (vectorizable_reduction): Make sure to
            set STMT_VINFO_REDUC_DEF for all live lanes in a SLP
            reduction.
            (vectorizable_live_operation): Do not pun to the SLP
            node representative for reduction epilogue generation.
    
            * gcc.dg/vect/pr107212-1.c: New testcase.
            * gcc.dg/vect/pr107212-2.c: Likewise.
    
    (cherry picked from commit ee467644c53ee2f7d633a8e1f53603feafab4351)

Diff:
---
 gcc/testsuite/gcc.dg/vect/pr107212-1.c | 27 +++++++++++++++++++++++++++
 gcc/testsuite/gcc.dg/vect/pr107212-2.c | 23 +++++++++++++++++++++++
 gcc/tree-vect-loop.c                   | 20 +++++++++++++-------
 3 files changed, 63 insertions(+), 7 deletions(-)

diff --git a/gcc/testsuite/gcc.dg/vect/pr107212-1.c b/gcc/testsuite/gcc.dg/vect/pr107212-1.c
new file mode 100644
index 00000000000..5343f9b6b23
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/vect/pr107212-1.c
@@ -0,0 +1,27 @@
+/* { dg-do run } */
+
+#include "tree-vect.h"
+
+int main()
+{
+  check_vect ();
+
+  unsigned int tab[6][2] = { {69, 73}, {36, 40}, {24, 16},
+        {16, 11}, {4, 5}, {3, 1} };
+
+  int sum_0 = 0;
+  int sum_1 = 0;
+
+  for(int t=0; t<6; t++) {
+      sum_0 += tab[t][0];
+      sum_1 += tab[t][1];
+  }
+
+  int x1 = (sum_0 < 100);
+  int x2 = (sum_0 > 200);
+
+  if (x1 || x2 || sum_1 != 146)
+    __builtin_abort ();
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.dg/vect/pr107212-2.c b/gcc/testsuite/gcc.dg/vect/pr107212-2.c
new file mode 100644
index 00000000000..109c2b991a6
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/vect/pr107212-2.c
@@ -0,0 +1,23 @@
+/* { dg-do run } */
+
+#include "tree-vect.h"
+
+int sum_1 = 0;
+
+int main()
+{
+  check_vect ();
+
+  unsigned int tab[6][2] = {{150, 0}, {0, 0}, {0, 0}, {0, 0}, {0, 0}, {0, 0}};
+  
+  int sum_0 = 0;
+  
+  for (int t = 0; t < 6; t++) {
+    sum_0 += tab[t][0];
+    sum_1 += tab[t][0];
+  }
+  
+  if (sum_0 < 100 || sum_0 > 200)
+    __builtin_abort();
+  return 0;
+}
diff --git a/gcc/tree-vect-loop.c b/gcc/tree-vect-loop.c
index 968087a9008..2c07fa38ff0 100644
--- a/gcc/tree-vect-loop.c
+++ b/gcc/tree-vect-loop.c
@@ -6589,10 +6589,20 @@ vectorizable_reduction (loop_vec_info loop_vinfo,
 	}
       if (!REDUC_GROUP_FIRST_ELEMENT (vdef))
 	only_slp_reduc_chain = false;
-      /* ???  For epilogue generation live members of the chain need
+      /* For epilogue generation live members of the chain need
          to point back to the PHI via their original stmt for
-	 info_for_reduction to work.  */
-      if (STMT_VINFO_LIVE_P (vdef))
+	 info_for_reduction to work.  For SLP we need to look at
+	 all lanes here - even though we only will vectorize from
+	 the SLP node with live lane zero the other live lanes also
+	 need to be identified as part of a reduction to be able
+	 to skip code generation for them.  */
+      if (slp_for_stmt_info)
+	{
+	  for (auto s : SLP_TREE_SCALAR_STMTS (slp_for_stmt_info))
+	    if (STMT_VINFO_LIVE_P (s))
+	      STMT_VINFO_REDUC_DEF (vect_orig_stmt (s)) = phi_info;
+	}
+      else if (STMT_VINFO_LIVE_P (vdef))
 	STMT_VINFO_REDUC_DEF (def) = phi_info;
       gassign *assign = dyn_cast <gassign *> (vdef->stmt);
       if (!assign)
@@ -8578,10 +8588,6 @@ vectorizable_live_operation (vec_info *vinfo,
 	     all involved stmts together.  */
 	  else if (slp_index != 0)
 	    return true;
-	  else
-	    /* For SLP reductions the meta-info is attached to
-	       the representative.  */
-	    stmt_info = SLP_TREE_REPRESENTATIVE (slp_node);
 	}
       stmt_vec_info reduc_info = info_for_reduction (loop_vinfo, stmt_info);
       gcc_assert (reduc_info->is_reduc_info);

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2023-01-24 15:22 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-01-24 15:22 [gcc r11-10480] tree-optimization/107212 - SLP reduction of reduction paths Richard Biener

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).