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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -force-streaming-compatible-sve < %s | FileCheck %s
target triple = "aarch64-unknown-linux-gnu"
define void @build_vector_7_inc1_v4i1(ptr %a) #0 {
; CHECK-LABEL: build_vector_7_inc1_v4i1:
; CHECK: // %bb.0:
; CHECK-NEXT: mov w8, #5
; CHECK-NEXT: strb w8, [x0]
; CHECK-NEXT: ret
store <4 x i1> <i1 true, i1 false, i1 true, i1 false>, ptr %a, align 1
ret void
}
define void @build_vector_7_inc1_v32i8(ptr %a) #0 {
; CHECK-LABEL: build_vector_7_inc1_v32i8:
; CHECK: // %bb.0:
; CHECK-NEXT: index z0.b, #0, #1
; CHECK-NEXT: mov z1.d, z0.d
; CHECK-NEXT: add z0.b, z0.b, #7 // =0x7
; CHECK-NEXT: add z1.b, z1.b, #23 // =0x17
; CHECK-NEXT: stp q0, q1, [x0]
; CHECK-NEXT: ret
store <32 x i8> <i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15, i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31, i8 32, i8 33, i8 34, i8 35, i8 36, i8 37, i8 38>, ptr %a, align 1
ret void
}
define void @build_vector_0_inc2_v16i16(ptr %a) #0 {
; CHECK-LABEL: build_vector_0_inc2_v16i16:
; CHECK: // %bb.0:
; CHECK-NEXT: index z0.h, #0, #2
; CHECK-NEXT: str q0, [x0]
; CHECK-NEXT: add z0.h, z0.h, #16 // =0x10
; CHECK-NEXT: str q0, [x0, #16]
; CHECK-NEXT: ret
store <16 x i16> <i16 0, i16 2, i16 4, i16 6, i16 8, i16 10, i16 12, i16 14, i16 16, i16 18, i16 20, i16 22, i16 24, i16 26, i16 28, i16 30>, ptr %a, align 2
ret void
}
; Negative const stride.
define void @build_vector_0_dec3_v8i32(ptr %a) #0 {
; CHECK-LABEL: build_vector_0_dec3_v8i32:
; CHECK: // %bb.0:
; CHECK-NEXT: index z0.s, #0, #-3
; CHECK-NEXT: mov z1.s, #-12 // =0xfffffffffffffff4
; CHECK-NEXT: add z1.s, z0.s, z1.s
; CHECK-NEXT: stp q0, q1, [x0]
; CHECK-NEXT: ret
store <8 x i32> <i32 0, i32 -3, i32 -6, i32 -9, i32 -12, i32 -15, i32 -18, i32 -21>, ptr %a, align 4
ret void
}
; Constant stride that's too big to be directly encoded into the index.
define void @build_vector_minus2_dec32_v4i64(ptr %a) #0 {
; CHECK-LABEL: build_vector_minus2_dec32_v4i64:
; CHECK: // %bb.0:
; CHECK-NEXT: mov x8, #-32
; CHECK-NEXT: mov z0.d, #-66 // =0xffffffffffffffbe
; CHECK-NEXT: mov z2.d, #-2 // =0xfffffffffffffffe
; CHECK-NEXT: index z1.d, #0, x8
; CHECK-NEXT: add z0.d, z1.d, z0.d
; CHECK-NEXT: add z1.d, z1.d, z2.d
; CHECK-NEXT: stp q1, q0, [x0]
; CHECK-NEXT: ret
store <4 x i64> <i64 -2, i64 -34, i64 -66, i64 -98>, ptr %a, align 8
ret void
}
; Constant but not a sequence.
define void @build_vector_no_stride_v4i64(ptr %a) #0 {
; CHECK-LABEL: build_vector_no_stride_v4i64:
; CHECK: // %bb.0:
; CHECK-NEXT: index z0.d, #1, #7
; CHECK-NEXT: index z1.d, #0, #4
; CHECK-NEXT: stp q1, q0, [x0]
; CHECK-NEXT: ret
store <4 x i64> <i64 0, i64 4, i64 1, i64 8>, ptr %a, align 8
ret void
}
define void @build_vector_0_inc2_v16f16(ptr %a) #0 {
; CHECK-LABEL: build_vector_0_inc2_v16f16:
; CHECK: // %bb.0:
; CHECK-NEXT: adrp x8, .LCPI6_0
; CHECK-NEXT: adrp x9, .LCPI6_1
; CHECK-NEXT: ldr q0, [x8, :lo12:.LCPI6_0]
; CHECK-NEXT: ldr q1, [x9, :lo12:.LCPI6_1]
; CHECK-NEXT: stp q1, q0, [x0]
; CHECK-NEXT: ret
store <16 x half> <half 0.0, half 2.0, half 4.0, half 6.0, half 8.0, half 10.0, half 12.0, half 14.0, half 16.0, half 18.0, half 20.0, half 22.0, half 24.0, half 26.0, half 28.0, half 30.0>, ptr %a, align 2
ret void
}
; Negative const stride.
define void @build_vector_0_dec3_v8f32(ptr %a) #0 {
; CHECK-LABEL: build_vector_0_dec3_v8f32:
; CHECK: // %bb.0:
; CHECK-NEXT: adrp x8, .LCPI7_0
; CHECK-NEXT: adrp x9, .LCPI7_1
; CHECK-NEXT: ldr q0, [x8, :lo12:.LCPI7_0]
; CHECK-NEXT: ldr q1, [x9, :lo12:.LCPI7_1]
; CHECK-NEXT: stp q1, q0, [x0]
; CHECK-NEXT: ret
store <8 x float> <float 0.0, float -3.0, float -6.0, float -9.0, float -12.0, float -15.0, float -18.0, float -21.0>, ptr %a, align 4
ret void
}
; Constant stride that's too big to be directly encoded into the index.
define void @build_vector_minus2_dec32_v4f64(ptr %a) #0 {
; CHECK-LABEL: build_vector_minus2_dec32_v4f64:
; CHECK: // %bb.0:
; CHECK-NEXT: adrp x8, .LCPI8_0
; CHECK-NEXT: adrp x9, .LCPI8_1
; CHECK-NEXT: ldr q0, [x8, :lo12:.LCPI8_0]
; CHECK-NEXT: ldr q1, [x9, :lo12:.LCPI8_1]
; CHECK-NEXT: stp q1, q0, [x0]
; CHECK-NEXT: ret
store <4 x double> <double -2.0, double -34.0, double -66.0, double -98.0>, ptr %a, align 8
ret void
}
; Constant but not a sequence.
define void @build_vector_no_stride_v4f64(ptr %a) #0 {
; CHECK-LABEL: build_vector_no_stride_v4f64:
; CHECK: // %bb.0:
; CHECK-NEXT: adrp x8, .LCPI9_0
; CHECK-NEXT: adrp x9, .LCPI9_1
; CHECK-NEXT: ldr q0, [x8, :lo12:.LCPI9_0]
; CHECK-NEXT: ldr q1, [x9, :lo12:.LCPI9_1]
; CHECK-NEXT: stp q1, q0, [x0]
; CHECK-NEXT: ret
store <4 x double> <double 0.0, double 4.0, double 1.0, double 8.0>, ptr %a, align 8
ret void
}
attributes #0 = { "target-features"="+sve" }
|
