xref: /llvm-project/llvm/test/CodeGen/AArch64/aarch64-vuzp.ll (revision 643c38333fc1b1e7e705e6e1035c595bbd95bc74)
1; RUN: llc -mtriple=aarch64-none-linux-gnu -mattr=+neon < %s | FileCheck %s
2
3declare <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8>, <16 x i8>)
4
5; CHECK-LABEL: fun1:
6; CHECK: uzp1 {{v[0-9]+}}.16b, {{v[0-9]+}}.16b, {{v[0-9]+}}.16b
7define i32 @fun1() {
8entry:
9  %vtbl1.i.1 = tail call <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8> <i8 0, i8 16, i8 19, i8 4, i8 -65, i8 -65, i8 -71, i8 -71, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, <16 x i8> undef)
10  %vuzp.i212.1 = shufflevector <16 x i8> %vtbl1.i.1, <16 x i8> %vtbl1.i.1, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
11  %scevgep = getelementptr <16 x i8>, ptr undef, i64 1
12  store <16 x i8> %vuzp.i212.1, ptr %scevgep, align 1
13  ret i32 undef
14}
15
16; CHECK-LABEL: fun2:
17; CHECK: uzp2 {{v[0-9]+}}.16b, {{v[0-9]+}}.16b, {{v[0-9]+}}.16b
18define i32 @fun2() {
19entry:
20  %vtbl1.i.1 = tail call <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8> <i8 0, i8 16, i8 19, i8 4, i8 -65, i8 -65, i8 -71, i8 -71, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, <16 x i8> undef)
21  %vuzp.i212.1 = shufflevector <16 x i8> %vtbl1.i.1, <16 x i8> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
22  %scevgep = getelementptr <8 x i8>, ptr undef, i64 1
23  store <8 x i8> %vuzp.i212.1, ptr %scevgep, align 1
24  ret i32 undef
25}
26
27; CHECK-LABEL: fun3:
28; CHECK-NOT: uzp1
29define i32 @fun3() {
30entry:
31  %vtbl1.i.1 = tail call <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8> <i8 0, i8 16, i8 19, i8 4, i8 -65, i8 -65, i8 -71, i8 -71, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, <16 x i8> undef)
32  %vuzp.i212.1 = shufflevector <16 x i8> %vtbl1.i.1, <16 x i8> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 15>
33  %scevgep = getelementptr <8 x i8>, ptr undef, i64 1
34  store <8 x i8> %vuzp.i212.1, ptr %scevgep, align 1
35  ret i32 undef
36}
37
38; CHECK-LABEL: fun4:
39; CHECK-NOT: uzp2
40define i32 @fun4() {
41entry:
42  %vtbl1.i.1 = tail call <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8> <i8 0, i8 16, i8 19, i8 4, i8 -65, i8 -65, i8 -71, i8 -71, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, <16 x i8> undef)
43  %vuzp.i212.1 = shufflevector <16 x i8> %vtbl1.i.1, <16 x i8> undef, <8 x i32> <i32 3, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
44  %scevgep = getelementptr <8 x i8>, ptr undef, i64 1
45  store <8 x i8> %vuzp.i212.1, ptr %scevgep, align 1
46  ret i32 undef
47}
48
49; CHECK-LABEL: pr36582:
50; Check that this does not ICE.
51define void @pr36582(ptr %p1, ptr %p2) {
52entry:
53  %wide.vec = load <8 x i8>, ptr %p1, align 1
54  %strided.vec = shufflevector <8 x i8> %wide.vec, <8 x i8> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
55  %y = zext <4 x i8> %strided.vec to <4 x i32>
56  store <4 x i32> %y, ptr %p2, align 4
57  ret void
58}
59
60; Check that this pattern is recognized as a VZIP and
61; that the vector blend transform does not scramble the pattern.
62; CHECK-LABEL: vzipNoBlend:
63; CHECK: zip1
64define <8 x i8> @vzipNoBlend(ptr %A, ptr %B) nounwind {
65  %t = load <8 x i8>, ptr %A
66  %vzip = shufflevector <8 x i8> %t, <8 x i8> <i8 0, i8 0, i8 0, i8 0, i8 undef, i8 undef, i8 undef, i8 undef>, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
67  ret <8 x i8> %vzip
68}
69
70; CHECK-LABEL: vzipNoBlendCommutted:
71; CHECK: zip1
72define <8 x i8> @vzipNoBlendCommutted(ptr %A, ptr %B) nounwind {
73  %t = load <8 x i8>, ptr %A
74  %vzip = shufflevector <8 x i8> <i8 0, i8 0, i8 0, i8 0, i8 undef, i8 undef, i8 undef, i8 undef>, <8 x i8> %t, <8 x i32> <i32 8, i32 0, i32 9, i32 1, i32 10, i32 2, i32 11, i32 3>
75  ret <8 x i8> %vzip
76}
77
78; CHECK-LABEL: vzipStillZExt:
79; CHECK: zip1
80define <8 x i8> @vzipStillZExt(ptr %A, ptr %B) nounwind {
81  %t = load <8 x i8>, ptr %A
82  %vzip = shufflevector <8 x i8> %t, <8 x i8> <i8 undef, i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <8 x i32> <i32 0, i32 9, i32 1, i32 9, i32 2, i32 9, i32 3, i32 9>
83  ret <8 x i8> %vzip
84}
85
86; CHECK-LABEL: vzipStillZExtCommutted:
87; CHECK: zip1
88define <8 x i8> @vzipStillZExtCommutted(ptr %A, ptr %B) nounwind {
89  %t = load <8 x i8>, ptr %A
90  %vzip = shufflevector <8 x i8> <i8 undef, i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <8 x i8> %t, <8 x i32> <i32 8, i32 1, i32 9, i32 1, i32 10, i32 1, i32 11, i32 1>
91  ret <8 x i8> %vzip
92}
93