| blob | 860d0ed29332c26b56b0b8adb52cf191c7a85c46 |
1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
2 ; RUN: opt -S --passes=slp-vectorizer -mtriple=x86_64-unknown-linux-gnu -mcpu=cascadelake < %s | FileCheck %s
4 define void @test(ptr %0, i32 %add651) {
5 ; CHECK-LABEL: define void @test(
6 ; CHECK-SAME: ptr [[TMP0:%.*]], i32 [[ADD651:%.*]]) #[[ATTR0:[0-9]+]] {
7 ; CHECK-NEXT: [[ENTRY:.*:]]
8 ; CHECK-NEXT: [[PREDPEL11:%.*]] = alloca [0 x [0 x [25 x i32]]], i32 0, align 16
9 ; CHECK-NEXT: [[ARRAYIDX469_6:%.*]] = getelementptr i8, ptr [[PREDPEL11]], i64 28
10 ; CHECK-NEXT: [[ARRAYIDX469_7:%.*]] = getelementptr i8, ptr [[PREDPEL11]], i64 32
11 ; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[PREDPEL11]], i64 36
12 ; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX469_7]], align 16
13 ; CHECK-NEXT: [[TMP3:%.*]] = load <2 x i32>, ptr [[ARRAYIDX469_6]], align 4
14 ; CHECK-NEXT: [[CONV470_7:%.*]] = trunc i32 [[TMP2]] to i16
15 ; CHECK-NEXT: store i16 [[CONV470_7]], ptr [[TMP0]], align 2
16 ; CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP0]], align 8
17 ; CHECK-NEXT: [[ARRAYIDX660:%.*]] = getelementptr i8, ptr [[TMP4]], i64 7800
18 ; CHECK-NEXT: [[ARRAYIDX689:%.*]] = getelementptr i8, ptr [[TMP4]], i64 7816
19 ; CHECK-NEXT: [[TMP5:%.*]] = load <2 x i32>, ptr [[TMP1]], align 4
20 ; CHECK-NEXT: [[TMP6:%.*]] = add <2 x i32> [[TMP3]], splat (i32 1)
21 ; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <2 x i32> [[TMP3]], <2 x i32> [[TMP5]], <2 x i32> <i32 1, i32 2>
22 ; CHECK-NEXT: [[TMP8:%.*]] = add <2 x i32> [[TMP6]], [[TMP7]]
23 ; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <2 x i32> [[TMP5]], <2 x i32> <i32 1, i32 poison>, <2 x i32> <i32 2, i32 1>
24 ; CHECK-NEXT: [[TMP10:%.*]] = add <2 x i32> [[TMP8]], [[TMP9]]
25 ; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[ADD651]], i32 0
26 ; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP2]], i32 1
27 ; CHECK-NEXT: [[TMP14:%.*]] = call <4 x i32> @llvm.vector.insert.v4i32.v2i32(<4 x i32> [[TMP13]], <2 x i32> [[TMP10]], i64 2)
28 ; CHECK-NEXT: [[TMP15:%.*]] = lshr <4 x i32> [[TMP14]], splat (i32 1)
29 ; CHECK-NEXT: [[SHR685:%.*]] = lshr i32 [[TMP2]], 1
30 ; CHECK-NEXT: [[TMP16:%.*]] = trunc <4 x i32> [[TMP15]] to <4 x i16>
31 ; CHECK-NEXT: [[CONV686:%.*]] = trunc i32 [[SHR685]] to i16
32 ; CHECK-NEXT: store i16 [[CONV686]], ptr [[ARRAYIDX689]], align 8
33 ; CHECK-NEXT: [[ARRAYIDX727:%.*]] = getelementptr i8, ptr [[TMP4]], i64 7818
34 ; CHECK-NEXT: [[TMP17:%.*]] = extractelement <4 x i16> [[TMP16]], i32 2
35 ; CHECK-NEXT: store i16 [[TMP17]], ptr [[ARRAYIDX727]], align 2
36 ; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i16> [[TMP16]], i32 3
37 ; CHECK-NEXT: store i16 [[TMP18]], ptr [[TMP4]], align 8
38 ; CHECK-NEXT: store <4 x i16> [[TMP16]], ptr [[ARRAYIDX660]], align 8
39 ; CHECK-NEXT: ret void
40 ;
41 entry:
42 %PredPel11 = alloca [0 x [0 x [25 x i32]]], i32 0, align 16
43 %arrayidx469.6 = getelementptr i8, ptr %PredPel11, i64 28
44 %1 = load i32, ptr %arrayidx469.6, align 4
45 %arrayidx469.7 = getelementptr i8, ptr %PredPel11, i64 32
46 %2 = load i32, ptr %arrayidx469.7, align 16
47 %conv470.7 = trunc i32 %2 to i16
48 store i16 %conv470.7, ptr %0, align 2
49 %3 = getelementptr i8, ptr %PredPel11, i64 36
50 %4 = getelementptr i8, ptr %PredPel11, i64 40
51 %5 = load ptr, ptr %0, align 8
52 %add6511 = add i32 %1, 1
53 %shr656 = lshr i32 %add651, 1
54 %conv657 = trunc i32 %shr656 to i16
55 %arrayidx660 = getelementptr i8, ptr %5, i64 7800
56 store i16 %conv657, ptr %arrayidx660, align 8
57 %shr685 = lshr i32 %2, 1
58 %conv686 = trunc i32 %shr685 to i16
59 %arrayidx689 = getelementptr i8, ptr %5, i64 7816
60 store i16 %conv686, ptr %arrayidx689, align 8
61 %arrayidx694 = getelementptr i8, ptr %5, i64 7802
62 store i16 %conv686, ptr %arrayidx694, align 2
63 %6 = load i32, ptr %3, align 4
64 %add716 = add i32 %add6511, %2
65 %add717 = add i32 %add716, 1
66 %shr718 = lshr i32 %add717, 1
67 %conv719 = trunc i32 %shr718 to i16
68 %arrayidx727 = getelementptr i8, ptr %5, i64 7818
69 store i16 %conv719, ptr %arrayidx727, align 2
70 %arrayidx731 = getelementptr i8, ptr %5, i64 7804
71 store i16 %conv719, ptr %arrayidx731, align 4
72 %7 = load i32, ptr %4, align 8
73 %add750 = add i32 %2, 1
74 %add753 = add i32 %add750, %6
75 %add754 = add i32 %add753, %7
76 %shr755 = lshr i32 %add754, 1
77 %conv756 = trunc i32 %shr755 to i16
78 store i16 %conv756, ptr %5, align 8
79 %arrayidx772 = getelementptr i8, ptr %5, i64 7806
80 store i16 %conv756, ptr %arrayidx772, align 2
81 ret void
82 }