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
|
; RUN: llc -mtriple=aarch64-linux-gnu -verify-machineinstrs -mattr=+sve -asm-verbose=0 < %s | FileCheck %s
;
; Unpredicated dup instruction (which is an alias for mov):
; * register + register,
; * register + immediate
;
define <vscale x 16 x i8> @dup_i8(i8 %b) {
; CHECK-LABEL: dup_i8:
; CHECK: mov z0.b, w0
; CHECK-NEXT: ret
%out = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 %b)
ret <vscale x 16 x i8> %out
}
define <vscale x 16 x i8> @dup_imm_i8() {
; CHECK-LABEL: dup_imm_i8:
; CHECK: mov z0.b, #16
; CHECK-NEXT: ret
%out = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 16)
ret <vscale x 16 x i8> %out
}
define <vscale x 8 x i16> @dup_i16(i16 %b) {
; CHECK-LABEL: dup_i16:
; CHECK: mov z0.h, w0
; CHECK-NEXT: ret
%out = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 %b)
ret <vscale x 8 x i16> %out
}
define <vscale x 8 x i16> @dup_imm_i16(i16 %b) {
; CHECK-LABEL: dup_imm_i16:
; CHECK: mov z0.h, #16
; CHECK-NEXT: ret
%out = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 16)
ret <vscale x 8 x i16> %out
}
define <vscale x 4 x i32> @dup_i32(i32 %b) {
; CHECK-LABEL: dup_i32:
; CHECK: mov z0.s, w0
; CHECK-NEXT: ret
%out = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 %b)
ret <vscale x 4 x i32> %out
}
define <vscale x 4 x i32> @dup_imm_i32(i32 %b) {
; CHECK-LABEL: dup_imm_i32:
; CHECK: mov z0.s, #16
; CHECK-NEXT: ret
%out = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 16)
ret <vscale x 4 x i32> %out
}
define <vscale x 2 x i64> @dup_i64(i64 %b) {
; CHECK-LABEL: dup_i64:
; CHECK: mov z0.d, x0
; CHECK-NEXT: ret
%out = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 %b)
ret <vscale x 2 x i64> %out
}
define <vscale x 2 x i64> @dup_imm_i64(i64 %b) {
; CHECK-LABEL: dup_imm_i64:
; CHECK: mov z0.d, #16
; CHECK-NEXT: ret
%out = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 16)
ret <vscale x 2 x i64> %out
}
define <vscale x 8 x half> @dup_f16(half %b) {
; CHECK-LABEL: dup_f16:
; CHECK: mov z0.h, h0
; CHECK-NEXT: ret
%out = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half %b)
ret <vscale x 8 x half> %out
}
define <vscale x 8 x bfloat> @dup_bf16(bfloat %b) #0 {
; CHECK-LABEL: dup_bf16:
; CHECK: mov z0.h, h0
; CHECK-NEXT: ret
%out = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat %b)
ret <vscale x 8 x bfloat> %out
}
define <vscale x 8 x half> @dup_imm_f16(half %b) {
; CHECK-LABEL: dup_imm_f16:
; CHECK: mov z0.h, #16.00000000
; CHECK-NEXT: ret
%out = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 16.)
ret <vscale x 8 x half> %out
}
define <vscale x 4 x float> @dup_f32(float %b) {
; CHECK-LABEL: dup_f32:
; CHECK: mov z0.s, s0
; CHECK-NEXT: ret
%out = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float %b)
ret <vscale x 4 x float> %out
}
define <vscale x 4 x float> @dup_imm_f32(float %b) {
; CHECK-LABEL: dup_imm_f32:
; CHECK: mov z0.s, #16.00000000
; CHECK-NEXT: ret
%out = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 16.)
ret <vscale x 4 x float> %out
}
define <vscale x 2 x double> @dup_f64(double %b) {
; CHECK-LABEL: dup_f64:
; CHECK: mov z0.d, d0
; CHECK-NEXT: ret
%out = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double %b)
ret <vscale x 2 x double> %out
}
define <vscale x 2 x double> @dup_imm_f64(double %b) {
; CHECK-LABEL: dup_imm_f64:
; CHECK: mov z0.d, #16.00000000
; CHECK-NEXT: ret
%out = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 16.)
ret <vscale x 2 x double> %out
}
define <vscale x 2 x float> @dup_fmov_imm_f32_2() {
; CHECK-LABEL: dup_fmov_imm_f32_2:
; CHECK: mov w8, #1109917696
; CHECK-NEXT: mov z0.s, w8
%out = tail call <vscale x 2 x float> @llvm.aarch64.sve.dup.x.nxv2f32(float 4.200000e+01)
ret <vscale x 2 x float> %out
}
define <vscale x 4 x float> @dup_fmov_imm_f32_4() {
; CHECK-LABEL: dup_fmov_imm_f32_4:
; CHECK: mov w8, #1109917696
; CHECK-NEXT: mov z0.s, w8
%out = tail call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 4.200000e+01)
ret <vscale x 4 x float> %out
}
define <vscale x 2 x double> @dup_fmov_imm_f64_2() {
; CHECK-LABEL: dup_fmov_imm_f64_2:
; CHECK: mov x8, #4631107791820423168
; CHECK-NEXT: mov z0.d, x8
%out = tail call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 4.200000e+01)
ret <vscale x 2 x double> %out
}
declare <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8( i8)
declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16)
declare <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32)
declare <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64)
declare <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half)
declare <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat)
declare <vscale x 2 x float> @llvm.aarch64.sve.dup.x.nxv2f32(float)
declare <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float)
declare <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double)
; +bf16 is required for the bfloat version.
attributes #0 = { "target-features"="+sve,+bf16" }
|
