1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py 2; RUN: %if x86-registered-target %{ opt < %s -passes=slp-vectorizer -S -mtriple=x86_64-unknown-linux-gnu | FileCheck %s %} 3; RUN: %if aarch64-registered-target %{ opt < %s -passes=slp-vectorizer -S -mtriple=aarch64-unknown-linux-gnu | FileCheck %s %} 4 5@b = common global [4 x i32] zeroinitializer, align 16 6@c = common global [4 x i32] zeroinitializer, align 16 7@d = common global [4 x i32] zeroinitializer, align 16 8@e = common global [4 x i32] zeroinitializer, align 16 9@a = common global [4 x i32] zeroinitializer, align 16 10@fb = common global [4 x float] zeroinitializer, align 16 11@fc = common global [4 x float] zeroinitializer, align 16 12@fa = common global [4 x float] zeroinitializer, align 16 13@fd = common global [4 x float] zeroinitializer, align 16 14 15define void @addsub() { 16; CHECK-LABEL: @addsub( 17; CHECK-NEXT: entry: 18; CHECK-NEXT: br label [[BB1:%.*]] 19; CHECK: bb1: 20; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i32>, ptr @b, align 16 21; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i32>, ptr @c, align 16 22; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[TMP0]], [[TMP1]] 23; CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr @d, align 16 24; CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr @e, align 16 25; CHECK-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[TMP3]], [[TMP4]] 26; CHECK-NEXT: [[TMP6:%.*]] = add nsw <4 x i32> [[TMP2]], [[TMP5]] 27; CHECK-NEXT: [[TMP7:%.*]] = sub nsw <4 x i32> [[TMP2]], [[TMP5]] 28; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <4 x i32> [[TMP6]], <4 x i32> [[TMP7]], <4 x i32> <i32 0, i32 5, i32 2, i32 7> 29; CHECK-NEXT: store <4 x i32> [[TMP8]], ptr @a, align 16 30; CHECK-NEXT: ret void 31; 32entry: 33 br label %bb1 34 35bb1: 36 %0 = load i32, ptr @b, align 16 37 %1 = load i32, ptr @c, align 16 38 %add = add nsw i32 %0, %1 39 %2 = load i32, ptr @d, align 16 40 %3 = load i32, ptr @e, align 16 41 %add1 = add nsw i32 %2, %3 42 %add2 = add nsw i32 %add, %add1 43 store i32 %add2, ptr @a, align 16 44 %4 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @b, i64 0, i64 1), align 4 45 %5 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @c, i64 0, i64 1), align 4 46 %add3 = add nsw i32 %4, %5 47 %6 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @d, i64 0, i64 1), align 4 48 %7 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @e, i64 0, i64 1), align 4 49 %add4 = add nsw i32 %6, %7 50 %sub = sub nsw i32 %add3, %add4 51 store i32 %sub, ptr getelementptr inbounds ([4 x i32], ptr @a, i64 0, i64 1), align 4 52 %8 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @b, i64 0, i64 2), align 8 53 %9 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @c, i64 0, i64 2), align 8 54 %add5 = add nsw i32 %8, %9 55 %10 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @d, i64 0, i64 2), align 8 56 %11 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @e, i64 0, i64 2), align 8 57 %add6 = add nsw i32 %10, %11 58 %add7 = add nsw i32 %add5, %add6 59 store i32 %add7, ptr getelementptr inbounds ([4 x i32], ptr @a, i64 0, i64 2), align 8 60 %12 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @b, i64 0, i64 3), align 4 61 %13 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @c, i64 0, i64 3), align 4 62 %add8 = add nsw i32 %12, %13 63 %14 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @d, i64 0, i64 3), align 4 64 %15 = load i32, ptr getelementptr inbounds ([4 x i32], ptr @e, i64 0, i64 3), align 4 65 %add9 = add nsw i32 %14, %15 66 %sub10 = sub nsw i32 %add8, %add9 67 store i32 %sub10, ptr getelementptr inbounds ([4 x i32], ptr @a, i64 0, i64 3), align 4 68 ret void 69} 70