1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt -mtriple=systemz-unknown -mcpu=z13 -passes=slp-vectorizer -S < %s | FileCheck %s
4 @bar = external global [4 x [4 x i32]], align 4
5 @dct_luma = external global [4 x [4 x i32]], align 4
10 ; CHECK-NEXT: [[ADD277:%.*]] = add nsw i32 undef, undef
11 ; CHECK-NEXT: store i32 [[ADD277]], ptr getelementptr inbounds ([4 x [4 x i32]], ptr @bar, i64 0, i64 3, i64 1), align 4
12 ; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr getelementptr inbounds ([4 x [4 x i32]], ptr @bar, i64 0, i64 3, i64 0), align 4
13 ; CHECK-NEXT: [[ARRAYIDX372:%.*]] = getelementptr inbounds [4 x [4 x i32]], ptr @dct_luma, i64 0, i64 3, i64 0
14 ; CHECK-NEXT: [[TMP1:%.*]] = load <2 x i32>, ptr getelementptr inbounds ([4 x [4 x i32]], ptr @bar, i64 0, i64 3, i64 2), align 4
15 ; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> <i32 poison, i32 undef, i32 poison, i32 poison>, i32 [[TMP0]], i32 0
16 ; CHECK-NEXT: [[TMP3:%.*]] = call <4 x i32> @llvm.vector.insert.v4i32.v2i32(<4 x i32> [[TMP2]], <2 x i32> [[TMP1]], i64 2)
17 ; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> <i32 undef, i32 poison, i32 undef, i32 undef>, i32 [[ADD277]], i32 1
18 ; CHECK-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[TMP3]], [[TMP4]]
19 ; CHECK-NEXT: [[TMP6:%.*]] = ashr <4 x i32> [[TMP5]], splat (i32 6)
20 ; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr [[ARRAYIDX372]], align 4
21 ; CHECK-NEXT: ret void
24 %add277 = add nsw i32 undef, undef
25 store i32 %add277, ptr getelementptr inbounds ([4 x [4 x i32]], ptr @bar, i64 0, i64 3, i64 1), align 4
26 %0 = load i32, ptr getelementptr inbounds ([4 x [4 x i32]], ptr @bar, i64 0, i64 3, i64 0), align 4
27 %sub355 = add nsw i32 undef, %0
28 %shr.i = ashr i32 %sub355, 6
29 %arrayidx372 = getelementptr inbounds [4 x [4 x i32]], ptr @dct_luma, i64 0, i64 3, i64 0
30 store i32 %shr.i, ptr %arrayidx372, align 4
31 %sub355.1 = add nsw i32 undef, %add277
32 %shr.i.1 = ashr i32 %sub355.1, 6
33 %arrayidx372.1 = getelementptr inbounds [4 x [4 x i32]], ptr @dct_luma, i64 0, i64 3, i64 1
34 store i32 %shr.i.1, ptr %arrayidx372.1, align 4
35 %1 = load i32, ptr getelementptr inbounds ([4 x [4 x i32]], ptr @bar, i64 0, i64 3, i64 2), align 4
36 %sub355.2 = add nsw i32 undef, %1
37 %shr.i.2 = ashr i32 %sub355.2, 6
38 %arrayidx372.2 = getelementptr inbounds [4 x [4 x i32]], ptr @dct_luma, i64 0, i64 3, i64 2
39 store i32 %shr.i.2, ptr %arrayidx372.2, align 4
40 %2 = load i32, ptr getelementptr inbounds ([4 x [4 x i32]], ptr @bar, i64 0, i64 3, i64 3), align 4
41 %sub355.3 = add nsw i32 undef, %2
42 %shr.i.3 = ashr i32 %sub355.3, 6
43 %arrayidx372.3 = getelementptr inbounds [4 x [4 x i32]], ptr @dct_luma, i64 0, i64 3, i64 3
44 store i32 %shr.i.3, ptr %arrayidx372.3, align 4