xref: /llvm-project/llvm/test/CodeGen/AArch64/fixed-vector-interleave.ll (revision 61510b51c33464a6bc15e4cf5b1ee07e2e0ec1c9)
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc -mtriple=aarch64-none-linux-gnu %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD
3; RUN: llc -mtriple=aarch64-none-linux-gnu -global-isel %s -o - 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI
4
5define <4 x half> @interleave2_v4f16(<2 x half> %vec0, <2 x half> %vec1) {
6; CHECK-LABEL: interleave2_v4f16:
7; CHECK:       // %bb.0:
8; CHECK-NEXT:    zip1 v0.4h, v0.4h, v1.4h
9; CHECK-NEXT:    ret
10  %retval = call <4 x half> @llvm.vector.interleave2.v4f16(<2 x half> %vec0, <2 x half> %vec1)
11  ret <4 x half> %retval
12}
13
14define <8 x half> @interleave2_v8f16(<4 x half> %vec0, <4 x half> %vec1) {
15; CHECK-SD-LABEL: interleave2_v8f16:
16; CHECK-SD:       // %bb.0:
17; CHECK-SD-NEXT:    // kill: def $d0 killed $d0 def $q0
18; CHECK-SD-NEXT:    // kill: def $d1 killed $d1 def $q1
19; CHECK-SD-NEXT:    adrp x8, .LCPI1_0
20; CHECK-SD-NEXT:    mov v0.d[1], v1.d[0]
21; CHECK-SD-NEXT:    ldr q1, [x8, :lo12:.LCPI1_0]
22; CHECK-SD-NEXT:    tbl v0.16b, { v0.16b }, v1.16b
23; CHECK-SD-NEXT:    ret
24;
25; CHECK-GI-LABEL: interleave2_v8f16:
26; CHECK-GI:       // %bb.0:
27; CHECK-GI-NEXT:    // kill: def $d0 killed $d0 def $q0
28; CHECK-GI-NEXT:    // kill: def $d1 killed $d1 def $q1
29; CHECK-GI-NEXT:    zip1 v0.8h, v0.8h, v1.8h
30; CHECK-GI-NEXT:    ret
31  %retval = call <8 x half> @llvm.vector.interleave2.v8f16(<4 x half> %vec0, <4 x half> %vec1)
32  ret <8 x half> %retval
33}
34
35define <16 x half> @interleave2_v16f16(<8 x half> %vec0, <8 x half> %vec1) {
36; CHECK-LABEL: interleave2_v16f16:
37; CHECK:       // %bb.0:
38; CHECK-NEXT:    zip1 v2.8h, v0.8h, v1.8h
39; CHECK-NEXT:    zip2 v1.8h, v0.8h, v1.8h
40; CHECK-NEXT:    mov v0.16b, v2.16b
41; CHECK-NEXT:    ret
42  %retval = call <16 x half> @llvm.vector.interleave2.v16f16(<8 x half> %vec0, <8 x half> %vec1)
43  ret <16 x half> %retval
44}
45
46define <4 x float> @interleave2_v4f32(<2 x float> %vec0, <2 x float> %vec1) {
47; CHECK-SD-LABEL: interleave2_v4f32:
48; CHECK-SD:       // %bb.0:
49; CHECK-SD-NEXT:    // kill: def $d0 killed $d0 def $q0
50; CHECK-SD-NEXT:    // kill: def $d1 killed $d1 def $q1
51; CHECK-SD-NEXT:    mov v0.d[1], v1.d[0]
52; CHECK-SD-NEXT:    rev64 v1.4s, v0.4s
53; CHECK-SD-NEXT:    uzp1 v0.4s, v0.4s, v1.4s
54; CHECK-SD-NEXT:    ret
55;
56; CHECK-GI-LABEL: interleave2_v4f32:
57; CHECK-GI:       // %bb.0:
58; CHECK-GI-NEXT:    // kill: def $d0 killed $d0 def $q0
59; CHECK-GI-NEXT:    // kill: def $d1 killed $d1 def $q1
60; CHECK-GI-NEXT:    zip1 v0.4s, v0.4s, v1.4s
61; CHECK-GI-NEXT:    ret
62  %retval = call <4 x float> @llvm.vector.interleave2.v4f32(<2 x float> %vec0, <2 x float> %vec1)
63  ret <4 x float> %retval
64}
65
66define <8 x float> @interleave2_v8f32(<4 x float> %vec0, <4 x float> %vec1) {
67; CHECK-LABEL: interleave2_v8f32:
68; CHECK:       // %bb.0:
69; CHECK-NEXT:    zip1 v2.4s, v0.4s, v1.4s
70; CHECK-NEXT:    zip2 v1.4s, v0.4s, v1.4s
71; CHECK-NEXT:    mov v0.16b, v2.16b
72; CHECK-NEXT:    ret
73  %retval = call <8 x float> @llvm.vector.interleave2.v8f32(<4 x float> %vec0, <4 x float> %vec1)
74  ret <8 x float> %retval
75}
76
77define <4 x double> @interleave2_v4f64(<2 x double> %vec0, <2 x double> %vec1) {
78; CHECK-LABEL: interleave2_v4f64:
79; CHECK:       // %bb.0:
80; CHECK-NEXT:    zip1 v2.2d, v0.2d, v1.2d
81; CHECK-NEXT:    zip2 v1.2d, v0.2d, v1.2d
82; CHECK-NEXT:    mov v0.16b, v2.16b
83; CHECK-NEXT:    ret
84  %retval = call <4 x double>@llvm.vector.interleave2.v4f64(<2 x double> %vec0, <2 x double> %vec1)
85  ret <4 x double> %retval
86}
87
88; Integers
89
90define <32 x i8> @interleave2_v32i8(<16 x i8> %vec0, <16 x i8> %vec1) {
91; CHECK-LABEL: interleave2_v32i8:
92; CHECK:       // %bb.0:
93; CHECK-NEXT:    zip1 v2.16b, v0.16b, v1.16b
94; CHECK-NEXT:    zip2 v1.16b, v0.16b, v1.16b
95; CHECK-NEXT:    mov v0.16b, v2.16b
96; CHECK-NEXT:    ret
97 %retval = call <32 x i8> @llvm.vector.interleave2.v32i8(<16 x i8> %vec0, <16 x i8> %vec1)
98  ret <32 x i8> %retval
99}
100
101define <16 x i16> @interleave2_v16i16(<8 x i16> %vec0, <8 x i16> %vec1) {
102; CHECK-LABEL: interleave2_v16i16:
103; CHECK:       // %bb.0:
104; CHECK-NEXT:    zip1 v2.8h, v0.8h, v1.8h
105; CHECK-NEXT:    zip2 v1.8h, v0.8h, v1.8h
106; CHECK-NEXT:    mov v0.16b, v2.16b
107; CHECK-NEXT:    ret
108  %retval = call <16 x i16> @llvm.vector.interleave2.v16i16(<8 x i16> %vec0, <8 x i16> %vec1)
109  ret <16 x i16> %retval
110}
111
112define <8 x i32> @interleave2_v8i32(<4 x i32> %vec0, <4 x i32> %vec1) {
113; CHECK-LABEL: interleave2_v8i32:
114; CHECK:       // %bb.0:
115; CHECK-NEXT:    zip1 v2.4s, v0.4s, v1.4s
116; CHECK-NEXT:    zip2 v1.4s, v0.4s, v1.4s
117; CHECK-NEXT:    mov v0.16b, v2.16b
118; CHECK-NEXT:    ret
119  %retval = call <8 x i32> @llvm.vector.interleave2.v8i32(<4 x i32> %vec0, <4 x i32> %vec1)
120  ret <8 x i32> %retval
121}
122
123define <4 x i64> @interleave2_v4i64(<2 x i64> %vec0, <2 x i64> %vec1) {
124; CHECK-LABEL: interleave2_v4i64:
125; CHECK:       // %bb.0:
126; CHECK-NEXT:    zip1 v2.2d, v0.2d, v1.2d
127; CHECK-NEXT:    zip2 v1.2d, v0.2d, v1.2d
128; CHECK-NEXT:    mov v0.16b, v2.16b
129; CHECK-NEXT:    ret
130  %retval = call <4 x i64> @llvm.vector.interleave2.v4i64(<2 x i64> %vec0, <2 x i64> %vec1)
131  ret <4 x i64> %retval
132}
133
134
135; Float declarations
136declare <4 x half> @llvm.vector.interleave2.v4f16(<2 x half>, <2 x half>)
137declare <8 x half> @llvm.vector.interleave2.v8f16(<4 x half>, <4 x half>)
138declare <16 x half> @llvm.vector.interleave2.v16f16(<8 x half>, <8 x half>)
139declare <4 x float> @llvm.vector.interleave2.v4f32(<2 x float>, <2 x float>)
140declare <8 x float> @llvm.vector.interleave2.v8f32(<4 x float>, <4 x float>)
141declare <4 x double> @llvm.vector.interleave2.v4f64(<2 x double>, <2 x double>)
142
143; Integer declarations
144declare <32 x i8> @llvm.vector.interleave2.v32i8(<16 x i8>, <16 x i8>)
145declare <16 x i16> @llvm.vector.interleave2.v16i16(<8 x i16>, <8 x i16>)
146declare <8 x i32> @llvm.vector.interleave2.v8i32(<4 x i32>, <4 x i32>)
147declare <4 x i64> @llvm.vector.interleave2.v4i64(<2 x i64>, <2 x i64>)
148
149