1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt < %s -passes=slp-vectorizer -S | FileCheck %s
3
4target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
5target triple = "x86_64-unknown-linux-gnu"
6
7@b = common global [4 x i32] zeroinitializer, align 16
8@c = common global [4 x i32] zeroinitializer, align 16
9@d = common global [4 x i32] zeroinitializer, align 16
10@e = common global [4 x i32] zeroinitializer, align 16
11@a = common global [4 x i32] zeroinitializer, align 16
12@fb = common global [4 x float] zeroinitializer, align 16
13@fc = common global [4 x float] zeroinitializer, align 16
14@fa = common global [4 x float] zeroinitializer, align 16
15@fd = common global [4 x float] zeroinitializer, align 16
16
17define void @addsub() {
18; CHECK-LABEL: @addsub(
19; CHECK-NEXT:  entry:
20; CHECK-NEXT:    br label [[BB1:%.*]]
21; CHECK:       bb1:
22; CHECK-NEXT:    [[TMP0:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @b to <4 x i32>*), align 16
23; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @c to <4 x i32>*), align 16
24; CHECK-NEXT:    [[TMP2:%.*]] = add nsw <4 x i32> [[TMP0]], [[TMP1]]
25; CHECK-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @d to <4 x i32>*), align 16
26; CHECK-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @e to <4 x i32>*), align 16
27; CHECK-NEXT:    [[TMP5:%.*]] = add nsw <4 x i32> [[TMP3]], [[TMP4]]
28; CHECK-NEXT:    [[TMP6:%.*]] = add nsw <4 x i32> [[TMP2]], [[TMP5]]
29; CHECK-NEXT:    [[TMP7:%.*]] = sub nsw <4 x i32> [[TMP2]], [[TMP5]]
30; CHECK-NEXT:    [[TMP8:%.*]] = shufflevector <4 x i32> [[TMP6]], <4 x i32> [[TMP7]], <4 x i32> <i32 0, i32 5, i32 2, i32 7>
31; CHECK-NEXT:    store <4 x i32> [[TMP8]], <4 x i32>* bitcast ([4 x i32]* @a to <4 x i32>*), align 16
32; CHECK-NEXT:    ret void
33;
34entry:
35  br label %bb1
36
37bb1:
38  %0 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i64 0, i64 0), align 16
39  %1 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i64 0, i64 0), align 16
40  %add = add nsw i32 %0, %1
41  %2 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i64 0, i64 0), align 16
42  %3 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i64 0, i64 0), align 16
43  %add1 = add nsw i32 %2, %3
44  %add2 = add nsw i32 %add, %add1
45  store i32 %add2, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i64 0, i64 0), align 16
46  %4 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i64 0, i64 1), align 4
47  %5 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i64 0, i64 1), align 4
48  %add3 = add nsw i32 %4, %5
49  %6 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i64 0, i64 1), align 4
50  %7 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i64 0, i64 1), align 4
51  %add4 = add nsw i32 %6, %7
52  %sub = sub nsw i32 %add3, %add4
53  store i32 %sub, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i64 0, i64 1), align 4
54  %8 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i64 0, i64 2), align 8
55  %9 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i64 0, i64 2), align 8
56  %add5 = add nsw i32 %8, %9
57  %10 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i64 0, i64 2), align 8
58  %11 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i64 0, i64 2), align 8
59  %add6 = add nsw i32 %10, %11
60  %add7 = add nsw i32 %add5, %add6
61  store i32 %add7, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i64 0, i64 2), align 8
62  %12 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i64 0, i64 3), align 4
63  %13 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i64 0, i64 3), align 4
64  %add8 = add nsw i32 %12, %13
65  %14 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i64 0, i64 3), align 4
66  %15 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i64 0, i64 3), align 4
67  %add9 = add nsw i32 %14, %15
68  %sub10 = sub nsw i32 %add8, %add9
69  store i32 %sub10, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i64 0, i64 3), align 4
70  ret void
71}
72