1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sme2 -force-streaming -verify-machineinstrs < %s | FileCheck %s
; == 8 to 64-bit elements ==
define <vscale x 16 x i8> @zip_x2_i8(<vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm) nounwind {
; CHECK-LABEL: zip_x2_i8:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.b, z3.b }, z0.b, z1.b
; CHECK-NEXT: add z0.b, z2.b, z0.b
; CHECK-NEXT: ret
%zip = call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.zip.x2.nxv16i8(<vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm)
%zip0 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %zip, 0
%add = add <vscale x 16 x i8> %zip0, %zn
ret <vscale x 16 x i8> %add
}
define <vscale x 8 x i16> @zip_x2_i16(<vscale x 8 x i16> %zn, <vscale x 8 x i16> %zm) nounwind {
; CHECK-LABEL: zip_x2_i16:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.h, z3.h }, z0.h, z1.h
; CHECK-NEXT: add z0.h, z2.h, z0.h
; CHECK-NEXT: ret
%zip = call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.zip.x2.nxv8i16(<vscale x 8 x i16> %zn, <vscale x 8 x i16> %zm)
%zip0 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %zip, 0
%add = add <vscale x 8 x i16> %zip0, %zn
ret <vscale x 8 x i16> %add
}
define <vscale x 8 x half> @zip_x2_f16(<vscale x 8 x half> %zn, <vscale x 8 x half> %zm) nounwind {
; CHECK-LABEL: zip_x2_f16:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.h, z3.h }, z0.h, z1.h
; CHECK-NEXT: fadd z0.h, z2.h, z0.h
; CHECK-NEXT: ret
%zip = call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.zip.x2.nxv8f16(<vscale x 8 x half> %zn, <vscale x 8 x half> %zm)
%zip0 = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } %zip, 0
%add = fadd <vscale x 8 x half> %zip0, %zn
ret <vscale x 8 x half> %add
}
define { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @zip_x2_bf16(<vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) nounwind {
; CHECK-LABEL: zip_x2_bf16:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.h, z1.h }, z0.h, z1.h
; CHECK-NEXT: ret
%zip = call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.zip.x2.nxv8bf16(<vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm)
ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } %zip
}
define <vscale x 4 x i32> @zip_x2_i32(<vscale x 4 x i32> %zn, <vscale x 4 x i32> %zm) nounwind {
; CHECK-LABEL: zip_x2_i32:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.s, z3.s }, z0.s, z1.s
; CHECK-NEXT: add z0.s, z2.s, z0.s
; CHECK-NEXT: ret
%zip = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.zip.x2.nxv4i32(<vscale x 4 x i32> %zn, <vscale x 4 x i32> %zm)
%zip0 = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } %zip, 0
%add = add <vscale x 4 x i32> %zip0, %zn
ret <vscale x 4 x i32> %add
}
define <vscale x 4 x float> @zip_x2_f32(<vscale x 4 x float> %zn, <vscale x 4 x float> %zm) nounwind {
; CHECK-LABEL: zip_x2_f32:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.s, z3.s }, z0.s, z1.s
; CHECK-NEXT: fadd z0.s, z2.s, z0.s
; CHECK-NEXT: ret
%zip = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.zip.x2.nxv4f32(<vscale x 4 x float> %zn, <vscale x 4 x float> %zm)
%zip0 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } %zip, 0
%add = fadd <vscale x 4 x float> %zip0, %zn
ret <vscale x 4 x float> %add
}
define <vscale x 2 x i64> @zip_x2_i64(<vscale x 2 x i64> %zn, <vscale x 2 x i64> %zm) nounwind {
; CHECK-LABEL: zip_x2_i64:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.d, z3.d }, z0.d, z1.d
; CHECK-NEXT: add z0.d, z2.d, z0.d
; CHECK-NEXT: ret
%zip = call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.zip.x2.nxv2i64(<vscale x 2 x i64> %zn, <vscale x 2 x i64> %zm)
%zip0 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } %zip, 0
%add = add <vscale x 2 x i64> %zip0, %zn
ret <vscale x 2 x i64> %add
}
define <vscale x 2 x double> @zip_x2_f64(<vscale x 2 x double> %zn, <vscale x 2 x double> %zm) nounwind {
; CHECK-LABEL: zip_x2_f64:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.d, z3.d }, z0.d, z1.d
; CHECK-NEXT: fadd z0.d, z2.d, z0.d
; CHECK-NEXT: ret
%zip = call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.zip.x2.nxv2f64(<vscale x 2 x double> %zn, <vscale x 2 x double> %zm)
%zip0 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %zip, 0
%add = fadd <vscale x 2 x double> %zip0, %zn
ret <vscale x 2 x double> %add
}
; == 128-bit elements ==
; NOTE: For the 128-bit case we only need to check the <vscale x 16 x i8> to
; ensure the tuple result starts at the correct register multiple. The other
; variants all test the same code path.
define <vscale x 16 x i8> @zipq_x2_i8(<vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm) nounwind {
; CHECK-LABEL: zipq_x2_i8:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z2.q, z3.q }, z0.q, z1.q
; CHECK-NEXT: add z0.b, z2.b, z0.b
; CHECK-NEXT: ret
%zip = call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.zipq.x2.nxv16i8(<vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm)
%zip0 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %zip, 0
%add = add <vscale x 16 x i8> %zip0, %zn
ret <vscale x 16 x i8> %add
}
define { <vscale x 8 x i16>, <vscale x 8 x i16> } @zipq_x2_i16(<vscale x 8 x i16> %zn, <vscale x 8 x i16> %zm) nounwind {
; CHECK-LABEL: zipq_x2_i16:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z0.q, z1.q
; CHECK-NEXT: ret
%res = call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.zipq.x2.nxv8i16(<vscale x 8 x i16> %zn, <vscale x 8 x i16> %zm)
ret { <vscale x 8 x i16>, <vscale x 8 x i16> } %res
}
define { <vscale x 8 x half>, <vscale x 8 x half> } @zipq_x2_f16(<vscale x 8 x half> %zn, <vscale x 8 x half> %zm) nounwind {
; CHECK-LABEL: zipq_x2_f16:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z0.q, z1.q
; CHECK-NEXT: ret
%res = call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.zipq.x2.nxv8f16(<vscale x 8 x half> %zn, <vscale x 8 x half> %zm)
ret { <vscale x 8 x half>, <vscale x 8 x half> } %res
}
define { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @zipq_x2_bf16(<vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) nounwind {
; CHECK-LABEL: zipq_x2_bf16:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z0.q, z1.q
; CHECK-NEXT: ret
%res = call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.zipq.x2.nxv8bf16(<vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm)
ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } %res
}
define { <vscale x 4 x i32>, <vscale x 4 x i32> } @zipq_x2_i32(<vscale x 4 x i32> %zn, <vscale x 4 x i32> %zm) nounwind {
; CHECK-LABEL: zipq_x2_i32:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z0.q, z1.q
; CHECK-NEXT: ret
%res = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.zipq.x2.nxv4i32(<vscale x 4 x i32> %zn, <vscale x 4 x i32> %zm)
ret { <vscale x 4 x i32>, <vscale x 4 x i32> } %res
}
define { <vscale x 4 x float>, <vscale x 4 x float> } @zipq_x2_f32(<vscale x 4 x float> %zn, <vscale x 4 x float> %zm) nounwind {
; CHECK-LABEL: zipq_x2_f32:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z0.q, z1.q
; CHECK-NEXT: ret
%res = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.zipq.x2.nxv4f32(<vscale x 4 x float> %zn, <vscale x 4 x float> %zm)
ret { <vscale x 4 x float>, <vscale x 4 x float> } %res
}
define { <vscale x 2 x i64>, <vscale x 2 x i64> } @zipq_x2_i64(<vscale x 2 x i64> %zn, <vscale x 2 x i64> %zm) nounwind {
; CHECK-LABEL: zipq_x2_i64:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z0.q, z1.q
; CHECK-NEXT: ret
%res = call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.zipq.x2.nxv2i64(<vscale x 2 x i64> %zn, <vscale x 2 x i64> %zm)
ret { <vscale x 2 x i64>, <vscale x 2 x i64> } %res
}
define { <vscale x 2 x double>, <vscale x 2 x double> } @zipq_x2_f64(<vscale x 2 x double> %zn, <vscale x 2 x double> %zm) nounwind {
; CHECK-LABEL: zipq_x2_f64:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z0.q, z1.q
; CHECK-NEXT: ret
%res = call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.zipq.x2.nxv2f64(<vscale x 2 x double> %zn, <vscale x 2 x double> %zm)
ret { <vscale x 2 x double>, <vscale x 2 x double> } %res
}
define { <vscale x 16 x i8>, <vscale x 16 x i8> } @zipq_x2_i8_not_tied(<vscale x 16 x i8> %unused, <vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm) nounwind {
; CHECK-LABEL: zipq_x2_i8_not_tied:
; CHECK: // %bb.0:
; CHECK-NEXT: zip { z0.q, z1.q }, z1.q, z2.q
; CHECK-NEXT: ret
%res = call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.zipq.x2.nxv16i8(<vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm)
ret { <vscale x 16 x i8>, <vscale x 16 x i8> } %res
}
; == 8 to 64-bit elements ==
declare { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.zip.x2.nxv16i8(<vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm)
declare { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.zip.x2.nxv8i16(<vscale x 8 x i16> %zn, <vscale x 8 x i16> %zm)
declare { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.zip.x2.nxv4i32(<vscale x 4 x i32> %zn, <vscale x 4 x i32> %zm)
declare { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.zip.x2.nxv2i64(<vscale x 2 x i64> %zn, <vscale x 2 x i64> %zm)
declare { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.zip.x2.nxv8f16(<vscale x 8 x half> %zn, <vscale x 8 x half> %zm)
declare { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.zip.x2.nxv8bf16(<vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm)
declare { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.zip.x2.nxv4f32(<vscale x 4 x float> %zn, <vscale x 4 x float> %zm)
declare { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.zip.x2.nxv2f64(<vscale x 2 x double> %zn, <vscale x 2 x double> %zm)
; == 128-bit elements ==
declare { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.zipq.x2.nxv16i8(<vscale x 16 x i8> %zn, <vscale x 16 x i8> %zm)
declare { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.zipq.x2.nxv8i16(<vscale x 8 x i16> %zn, <vscale x 8 x i16> %zm)
declare { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.zipq.x2.nxv4i32(<vscale x 4 x i32> %zn, <vscale x 4 x i32> %zm)
declare { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.zipq.x2.nxv2i64(<vscale x 2 x i64> %zn, <vscale x 2 x i64> %zm)
declare { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.zipq.x2.nxv8f16(<vscale x 8 x half> %zn, <vscale x 8 x half> %zm)
declare { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.zipq.x2.nxv8bf16(<vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm)
declare { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.zipq.x2.nxv4f32(<vscale x 4 x float> %zn, <vscale x 4 x float> %zm)
declare { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.zipq.x2.nxv2f64(<vscale x 2 x double> %zn, <vscale x 2 x double> %zm)
|
