summaryrefslogtreecommitdiff
path: root/0007-Vect-Enable-skipping-vectorization-on-reduction-chai.patch
blob: ffe5327035d3daf45d0d265b2e9dd850100d95f9 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
From 07033bcc5b9e4c03846cd84b4587cd493fcf7d53 Mon Sep 17 00:00:00 2001
From: zhoukaipeng <zhoukaipeng3@huawei.com>
Date: Wed, 14 Jul 2021 11:24:06 +0800
Subject: [PATCH 07/13] [Vect] Enable skipping vectorization on reduction
 chains

Sometimes either vectorization on reduction chains or reductions is
possible. But the latter is better. The option "-ftree-vect-analyze
-slp-group" skips the former.

diff --git a/gcc/common.opt b/gcc/common.opt
index 8eb05570418..55d4eb5a351 100644
--- a/gcc/common.opt
+++ b/gcc/common.opt
@@ -2968,6 +2968,10 @@ ftree-slp-vectorize
 Common Report Var(flag_tree_slp_vectorize) Optimization EnabledBy(ftree-vectorize)
 Enable basic block vectorization (SLP) on trees.
 
+ftree-vect-analyze-slp-group
+Common Report Var(flag_tree_slp_group) Init(0)
+Disable SLP vectorization for reduction chain on tree.
+
 fvect-cost-model=
 Common Joined RejectNegative Enum(vect_cost_model) Var(flag_vect_cost_model) Init(VECT_COST_MODEL_DEFAULT) Optimization
 -fvect-cost-model=[unlimited|dynamic|cheap]	Specifies the cost model for vectorization.
diff --git a/gcc/testsuite/gcc.dg/vect/vect-reduc-12.c b/gcc/testsuite/gcc.dg/vect/vect-reduc-12.c
new file mode 100644
index 00000000000..913f1ef28df
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/vect/vect-reduc-12.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -funsafe-math-optimizations -fno-tree-reassoc -ftree-vect-analyze-slp-group" } */
+void f(double *a, double *res, double m) {
+  double res1, res0;
+  res1 = 0;
+  res0 = 0;
+  for (int i = 0; i < 1000; i+=8) {
+    res0 += a[i] * m;
+    res1 += a[i+1] * m;
+    res0 += a[i+2] * m;
+    res1 += a[i+3] * m;
+    res0 += a[i+4] * m;
+    res1 += a[i+5] * m;
+    res0 += a[i+6] * m;
+    res1 += a[i+7] * m;
+  }
+  res[0] += res0;
+  res[1] += res1;
+}
+/* { dg-final { scan-tree-dump "vectorized 1 loops" "vect" } } */
diff --git a/gcc/tree-vect-slp.c b/gcc/tree-vect-slp.c
index adc579ff544..476b3237054 100644
--- a/gcc/tree-vect-slp.c
+++ b/gcc/tree-vect-slp.c
@@ -2480,7 +2480,8 @@ vect_analyze_slp (vec_info *vinfo, unsigned max_tree_size)
 	{
 	  /* Find SLP sequences starting from reduction chains.  */
 	  FOR_EACH_VEC_ELT (loop_vinfo->reduction_chains, i, first_element)
-	    if (! vect_analyze_slp_instance (vinfo, bst_map, first_element,
+	    if (flag_tree_slp_group
+		|| ! vect_analyze_slp_instance (vinfo, bst_map, first_element,
 					     max_tree_size))
 	      {
 		/* Dissolve reduction chain group.  */
-- 
2.21.0.windows.1