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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
; RUN: llc -mtriple=aarch64 < %s | FileCheck %s
define i32 @lower_lshr(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c, <4 x i32> %d, <4 x i32> %e, <4 x i32> %f, <4 x i32> %g, <4 x i32> %h) {
; CHECK-LABEL: lower_lshr:
; CHECK: // %bb.0:
; CHECK-NEXT: addv s0, v0.4s
; CHECK-NEXT: addv s1, v1.4s
; CHECK-NEXT: addv s4, v4.4s
; CHECK-NEXT: addv s5, v5.4s
; CHECK-NEXT: addv s2, v2.4s
; CHECK-NEXT: addv s6, v6.4s
; CHECK-NEXT: mov v0.s[1], v1.s[0]
; CHECK-NEXT: addv s1, v3.4s
; CHECK-NEXT: addv s3, v7.4s
; CHECK-NEXT: mov v4.s[1], v5.s[0]
; CHECK-NEXT: mov v0.s[2], v2.s[0]
; CHECK-NEXT: mov v4.s[2], v6.s[0]
; CHECK-NEXT: mov v0.s[3], v1.s[0]
; CHECK-NEXT: mov v4.s[3], v3.s[0]
; CHECK-NEXT: xtn v1.4h, v0.4s
; CHECK-NEXT: shrn v0.4h, v0.4s, #16
; CHECK-NEXT: xtn v2.4h, v4.4s
; CHECK-NEXT: shrn v3.4h, v4.4s, #16
; CHECK-NEXT: uhadd v0.4h, v1.4h, v0.4h
; CHECK-NEXT: uhadd v1.4h, v2.4h, v3.4h
; CHECK-NEXT: mov v0.d[1], v1.d[0]
; CHECK-NEXT: uaddlv s0, v0.8h
; CHECK-NEXT: fmov w0, s0
; CHECK-NEXT: ret
%l87 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a)
%l174 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %b)
%l257 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %c)
%l340 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %d)
%l427 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %e)
%l514 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %f)
%l597 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %g)
%l680 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %h)
%l681 = insertelement <8 x i32> poison, i32 %l87, i32 0
%l682 = insertelement <8 x i32> %l681, i32 %l174, i32 1
%l683 = insertelement <8 x i32> %l682, i32 %l257, i32 2
%l684 = insertelement <8 x i32> %l683, i32 %l340, i32 3
%l685 = insertelement <8 x i32> %l684, i32 %l427, i32 4
%l686 = insertelement <8 x i32> %l685, i32 %l514, i32 5
%l687 = insertelement <8 x i32> %l686, i32 %l597, i32 6
%l688 = insertelement <8 x i32> %l687, i32 %l680, i32 7
%l689 = and <8 x i32> %l688, <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>
%l690 = lshr <8 x i32> %l688, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16>
%l691 = add nuw nsw <8 x i32> %l689, %l690
%l692 = lshr <8 x i32> %l691, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%l693 = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %l692)
ret i32 %l693
}
declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>)
declare i32 @llvm.vector.reduce.add.v8i32(<8 x i32>)
define <16 x i8> @lower_trunc_16xi8(i16 %a, i16 %b, i16 %c, i16 %d, i16 %e, i16 %f, i16 %g, i16 %h, i16 %i, i16 %j, i16 %k, i16 %l, i16 %m, i16 %n, i16 %o, i16 %p) {
; CHECK-LABEL: lower_trunc_16xi8:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov s0, w0
; CHECK-NEXT: add x8, sp, #56
; CHECK-NEXT: ld1r { v1.8h }, [x8]
; CHECK-NEXT: mov v0.h[1], w1
; CHECK-NEXT: add v3.8h, v1.8h, v1.8h
; CHECK-NEXT: mov v0.h[2], w2
; CHECK-NEXT: mov v0.h[3], w3
; CHECK-NEXT: mov v0.h[4], w4
; CHECK-NEXT: mov v0.h[5], w5
; CHECK-NEXT: mov v0.h[6], w6
; CHECK-NEXT: add v2.8h, v0.8h, v0.8h
; CHECK-NEXT: uzp1 v0.16b, v0.16b, v1.16b
; CHECK-NEXT: uzp1 v1.16b, v2.16b, v3.16b
; CHECK-NEXT: eor v0.16b, v0.16b, v1.16b
; CHECK-NEXT: ret
%a1 = insertelement <16 x i16> poison, i16 %a, i16 0
%b1 = insertelement <16 x i16> %a1, i16 %b, i16 1
%c1 = insertelement <16 x i16> %b1, i16 %c, i16 2
%d1 = insertelement <16 x i16> %c1, i16 %d, i16 3
%e1 = insertelement <16 x i16> %d1, i16 %e, i16 4
%f1 = insertelement <16 x i16> %e1, i16 %f, i16 5
%g1 = insertelement <16 x i16> %f1, i16 %g, i16 6
%h1 = insertelement <16 x i16> %g1, i16 %h, i16 7
%i1 = insertelement <16 x i16> %f1, i16 %i, i16 8
%j1 = insertelement <16 x i16> %g1, i16 %j, i16 9
%k1 = insertelement <16 x i16> %f1, i16 %k, i16 10
%l1 = insertelement <16 x i16> %g1, i16 %l, i16 11
%m1 = insertelement <16 x i16> %f1, i16 %m, i16 12
%n1 = insertelement <16 x i16> %g1, i16 %n, i16 13
%o1 = insertelement <16 x i16> %f1, i16 %o, i16 14
%p1 = insertelement <16 x i16> %g1, i16 %p, i16 15
%t = trunc <16 x i16> %p1 to <16 x i8>
%s = add <16 x i16> %p1, %p1
%t2 = trunc <16 x i16> %s to <16 x i8>
%pro = xor <16 x i8> %t, %t2
ret <16 x i8> %pro
}
define <8 x i16> @lower_trunc_8xi16(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e, i32 %f, i32 %g, i32 %h) {
; CHECK-LABEL: lower_trunc_8xi16:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov s0, w4
; CHECK-NEXT: fmov s1, w0
; CHECK-NEXT: mov v0.s[1], w5
; CHECK-NEXT: mov v1.s[1], w1
; CHECK-NEXT: mov v0.s[2], w6
; CHECK-NEXT: mov v1.s[2], w2
; CHECK-NEXT: mov v0.s[3], w7
; CHECK-NEXT: mov v1.s[3], w3
; CHECK-NEXT: add v2.4s, v0.4s, v0.4s
; CHECK-NEXT: add v3.4s, v1.4s, v1.4s
; CHECK-NEXT: uzp1 v0.8h, v1.8h, v0.8h
; CHECK-NEXT: uzp1 v1.8h, v3.8h, v2.8h
; CHECK-NEXT: eor v0.16b, v0.16b, v1.16b
; CHECK-NEXT: ret
%a1 = insertelement <8 x i32> poison, i32 %a, i32 0
%b1 = insertelement <8 x i32> %a1, i32 %b, i32 1
%c1 = insertelement <8 x i32> %b1, i32 %c, i32 2
%d1 = insertelement <8 x i32> %c1, i32 %d, i32 3
%e1 = insertelement <8 x i32> %d1, i32 %e, i32 4
%f1 = insertelement <8 x i32> %e1, i32 %f, i32 5
%g1 = insertelement <8 x i32> %f1, i32 %g, i32 6
%h1 = insertelement <8 x i32> %g1, i32 %h, i32 7
%t = trunc <8 x i32> %h1 to <8 x i16>
%s = add <8 x i32> %h1, %h1
%t2 = trunc <8 x i32> %s to <8 x i16>
%o = xor <8 x i16> %t, %t2
ret <8 x i16> %o
}
define <4 x i32> @lower_trunc_4xi32(i64 %a, i64 %b, i64 %c, i64 %d) {
; CHECK-LABEL: lower_trunc_4xi32:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov d0, x2
; CHECK-NEXT: fmov d1, x0
; CHECK-NEXT: mov v0.d[1], x3
; CHECK-NEXT: mov v1.d[1], x1
; CHECK-NEXT: add v2.2d, v0.2d, v0.2d
; CHECK-NEXT: add v3.2d, v1.2d, v1.2d
; CHECK-NEXT: uzp1 v0.4s, v1.4s, v0.4s
; CHECK-NEXT: uzp1 v1.4s, v3.4s, v2.4s
; CHECK-NEXT: eor v0.16b, v0.16b, v1.16b
; CHECK-NEXT: ret
%a1 = insertelement <4 x i64> poison, i64 %a, i64 0
%b1 = insertelement <4 x i64> %a1, i64 %b, i64 1
%c1 = insertelement <4 x i64> %b1, i64 %c, i64 2
%d1 = insertelement <4 x i64> %c1, i64 %d, i64 3
%t = trunc <4 x i64> %d1 to <4 x i32>
%s = add <4 x i64> %d1, %d1
%t2 = trunc <4 x i64> %s to <4 x i32>
%o = xor <4 x i32> %t, %t2
ret <4 x i32> %o
}
define <8 x i32> @lower_trunc_8xi32(i64 %a, i64 %b, i64 %c, i64 %d, i64 %e, i64 %f, i64 %g, i64 %h) {
; CHECK-LABEL: lower_trunc_8xi32:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov d0, x2
; CHECK-NEXT: fmov d1, x0
; CHECK-NEXT: fmov d2, x6
; CHECK-NEXT: fmov d3, x4
; CHECK-NEXT: mov v0.d[1], x3
; CHECK-NEXT: mov v1.d[1], x1
; CHECK-NEXT: mov v2.d[1], x7
; CHECK-NEXT: mov v3.d[1], x5
; CHECK-NEXT: add v4.2d, v0.2d, v0.2d
; CHECK-NEXT: add v5.2d, v1.2d, v1.2d
; CHECK-NEXT: add v6.2d, v2.2d, v2.2d
; CHECK-NEXT: add v7.2d, v3.2d, v3.2d
; CHECK-NEXT: uzp1 v2.4s, v3.4s, v2.4s
; CHECK-NEXT: uzp1 v0.4s, v1.4s, v0.4s
; CHECK-NEXT: uzp1 v3.4s, v5.4s, v4.4s
; CHECK-NEXT: uzp1 v1.4s, v7.4s, v6.4s
; CHECK-NEXT: eor v0.16b, v0.16b, v3.16b
; CHECK-NEXT: eor v1.16b, v2.16b, v1.16b
; CHECK-NEXT: ret
%a1 = insertelement <8 x i64> poison, i64 %a, i64 0
%b1 = insertelement <8 x i64> %a1, i64 %b, i64 1
%c1 = insertelement <8 x i64> %b1, i64 %c, i64 2
%d1 = insertelement <8 x i64> %c1, i64 %d, i64 3
%e1 = insertelement <8 x i64> %d1, i64 %e, i64 4
%f1 = insertelement <8 x i64> %e1, i64 %f, i64 5
%g1 = insertelement <8 x i64> %f1, i64 %g, i64 6
%h1 = insertelement <8 x i64> %g1, i64 %h, i64 7
%t = trunc <8 x i64> %h1 to <8 x i32>
%s = add <8 x i64> %h1, %h1
%t2 = trunc <8 x i64> %s to <8 x i32>
%o = xor <8 x i32> %t, %t2
ret <8 x i32> %o
}
|
