From: Richard Sandiford <richard.sandiford@arm.com>
Date: Thu, 9 Aug 2018 16:03:25 +0000 (+0000)
Subject: Allow inner-loop reductions with variable-length vectors
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=b8d5e148f669d56ccfd294dc396daba607237cdb;p=gcc.git

Allow inner-loop reductions with variable-length vectors

While working on PR 86871, I noticed we were being overly restrictive
when handling variable-length vectors.  For:

  for (i : ...)
    {
      res = ...;
      for (j : ...)
        res op= ...;
      a[i] = res;
    }

we don't need a reduction operation (although we do for double
reductions like:

  res = ...;
  for (i : ...)
    for (j : ...)
      res op= ...;
  a[i] = res;

which must still be rejected).

2018-08-08  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
	* tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
	reductions for variable-length vectors.

gcc/testsuite/
	* gcc.target/aarch64/sve/reduc_8.c: New test.

From-SVN: r263451
---

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index dd088bb46a8..839dfc3c2c2 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,8 @@
+2018-08-09  Richard Sandiford  <richard.sandiford@arm.com>
+
+	* tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
+	reductions for variable-length vectors.
+
 2018-08-09  David Malcolm  <dmalcolm@redhat.com>
 
 	PR other/84889
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 3da9f3baa36..8ea8530243d 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,7 @@
+2018-08-09  Richard Sandiford  <richard.sandiford@arm.com>
+
+	* gcc.target/aarch64/sve/reduc_8.c: New test.
+
 2018-08-09  David Malcolm  <dmalcolm@redhat.com>
 
 	PR other/84889
diff --git a/gcc/testsuite/gcc.target/aarch64/sve/reduc_8.c b/gcc/testsuite/gcc.target/aarch64/sve/reduc_8.c
new file mode 100644
index 00000000000..3913b8848c0
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/sve/reduc_8.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -ftree-vectorize" } */
+
+int
+reduc (int *restrict a, int *restrict b, int *restrict c)
+{
+  for (int i = 0; i < 100; ++i)
+    {
+      int res = 0;
+      for (int j = 0; j < 100; ++j)
+	if (b[i + j] != 0)
+	  res = c[i + j];
+      a[i] = res;
+    }
+}
+
+/* { dg-final { scan-assembler-times {\tcmpne\tp[0-9]+\.s, } 1 } } */
+/* We ought to use the CMPNE result for the SEL too.  */
+/* { dg-final { scan-assembler-not {\tcmpeq\tp[0-9]+\.s, } { xfail *-*-* } } } */
+/* { dg-final { scan-assembler-times {\tsel\tz[0-9]+\.s, } 1 } } */
diff --git a/gcc/tree-vect-loop.c b/gcc/tree-vect-loop.c
index 0669f62c960..c167aec326e 100644
--- a/gcc/tree-vect-loop.c
+++ b/gcc/tree-vect-loop.c
@@ -6714,6 +6714,7 @@ vectorizable_reduction (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi,
     }
 
   if (reduction_type != EXTRACT_LAST_REDUCTION
+      && (!nested_cycle || double_reduc)
       && reduc_fn == IFN_LAST
       && !nunits_out.is_constant ())
     {