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
209
210
211
212
213
214
215
216
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s | FileCheck %s
target triple = "aarch64-unknown-linux-gnu"
; Check that we don't try and merge uunpklo/uzp1 with a load or store if we
; would end up creating a predicate that would be too large for the max VL.
; UUNPKLO + Load
define <vscale x 8 x i16> @uunpklo_i8_valid(ptr %b) #0 {
; CHECK-LABEL: uunpklo_i8_valid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.h, vl64
; CHECK-NEXT: ld1b { z0.h }, p0/z, [x0]
; CHECK-NEXT: ret
%mask = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 11)
%load = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8(ptr %b, i32 2, <vscale x 16 x i1> %mask, <vscale x 16 x i8> undef)
%uzp = call <vscale x 8 x i16> @llvm.aarch64.sve.uunpklo.nxv8i16(<vscale x 16 x i8> %load)
ret <vscale x 8 x i16> %uzp
}
define <vscale x 8 x i16> @uunpklo_i8_invalid(ptr %b) #0 {
; CHECK-LABEL: uunpklo_i8_invalid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.b, vl128
; CHECK-NEXT: ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT: uunpklo z0.h, z0.b
; CHECK-NEXT: ret
%mask = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 12)
%load = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8(ptr %b, i32 2, <vscale x 16 x i1> %mask, <vscale x 16 x i8> undef)
%uzp = call <vscale x 8 x i16> @llvm.aarch64.sve.uunpklo.nxv8i16(<vscale x 16 x i8> %load)
ret <vscale x 8 x i16> %uzp
}
define <vscale x 4 x i32> @uunpklo_i16_valid(ptr %b) #0 {
; CHECK-LABEL: uunpklo_i16_valid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.s, vl32
; CHECK-NEXT: ld1h { z0.s }, p0/z, [x0]
; CHECK-NEXT: ret
%mask = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 10)
%load = call <vscale x 8 x i16> @llvm.masked.load.nxv8i16(ptr %b, i32 2, <vscale x 8 x i1> %mask, <vscale x 8 x i16> undef)
%uzp = call <vscale x 4 x i32> @llvm.aarch64.sve.uunpklo.nxv4i32(<vscale x 8 x i16> %load)
ret <vscale x 4 x i32> %uzp
}
define <vscale x 4 x i32> @uunpklo_i16_invalid(ptr %b) #0 {
; CHECK-LABEL: uunpklo_i16_invalid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.h, vl64
; CHECK-NEXT: ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT: uunpklo z0.s, z0.h
; CHECK-NEXT: ret
%mask = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 11)
%load = call <vscale x 8 x i16> @llvm.masked.load.nxv8i16(ptr %b, i32 2, <vscale x 8 x i1> %mask, <vscale x 8 x i16> undef)
%uzp = call <vscale x 4 x i32> @llvm.aarch64.sve.uunpklo.nxv4i32(<vscale x 8 x i16> %load)
ret <vscale x 4 x i32> %uzp
}
define <vscale x 2 x i64> @uunpklo_i32_valid(ptr %b) #0 {
; CHECK-LABEL: uunpklo_i32_valid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.d, vl16
; CHECK-NEXT: ld1w { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%mask = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 9)
%load = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32(ptr %b, i32 2, <vscale x 4 x i1> %mask, <vscale x 4 x i32> undef)
%uzp = call <vscale x 2 x i64> @llvm.aarch64.sve.uunpklo.nxv2i64(<vscale x 4 x i32> %load)
ret <vscale x 2 x i64> %uzp
}
define <vscale x 2 x i64> @uunpklo_i32_invalid(ptr %b) #0 {
; CHECK-LABEL: uunpklo_i32_invalid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.s, vl32
; CHECK-NEXT: ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT: uunpklo z0.d, z0.s
; CHECK-NEXT: ret
%mask = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 10)
%load = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32(ptr %b, i32 2, <vscale x 4 x i1> %mask, <vscale x 4 x i32> undef)
%uzp = call <vscale x 2 x i64> @llvm.aarch64.sve.uunpklo.nxv2i64(<vscale x 4 x i32> %load)
ret <vscale x 2 x i64> %uzp
}
define <vscale x 2 x i64> @uunpklo_invalid_all(ptr %b) #0 {
; CHECK-LABEL: uunpklo_invalid_all:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.s
; CHECK-NEXT: ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT: uunpklo z0.d, z0.s
; CHECK-NEXT: ret
%mask = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%load = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32(ptr %b, i32 2, <vscale x 4 x i1> %mask, <vscale x 4 x i32> undef)
%uzp = call <vscale x 2 x i64> @llvm.aarch64.sve.uunpklo.nxv2i64(<vscale x 4 x i32> %load)
ret <vscale x 2 x i64> %uzp
}
; UZP1 + Store
define void @uzp1_i8_valid(<vscale x 8 x i16> %a, ptr %b) #0 {
; CHECK-LABEL: uzp1_i8_valid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.h, vl64
; CHECK-NEXT: st1b { z0.h }, p0, [x0]
; CHECK-NEXT: ret
%a.bc = bitcast <vscale x 8 x i16> %a to <vscale x 16 x i8>
%uzp = call <vscale x 16 x i8> @llvm.aarch64.sve.uzp1.nxv16i8(<vscale x 16 x i8> %a.bc, <vscale x 16 x i8> %a.bc)
%mask = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 11)
call void @llvm.masked.store.nxv16i8(<vscale x 16 x i8> %uzp, ptr %b, i32 2, <vscale x 16 x i1> %mask)
ret void
}
define void @uzp1_i8_invalid(<vscale x 8 x i16> %a, ptr %b) #0 {
; CHECK-LABEL: uzp1_i8_invalid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.b, vl128
; CHECK-NEXT: uzp1 z0.b, z0.b, z0.b
; CHECK-NEXT: st1b { z0.b }, p0, [x0]
; CHECK-NEXT: ret
%a.bc = bitcast <vscale x 8 x i16> %a to <vscale x 16 x i8>
%uzp = call <vscale x 16 x i8> @llvm.aarch64.sve.uzp1.nxv16i8(<vscale x 16 x i8> %a.bc, <vscale x 16 x i8> %a.bc)
%mask = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 12)
call void @llvm.masked.store.nxv16i8(<vscale x 16 x i8> %uzp, ptr %b, i32 2, <vscale x 16 x i1> %mask)
ret void
}
define void @uzp1_i16_valid(<vscale x 4 x i32> %a, ptr %b) #0 {
; CHECK-LABEL: uzp1_i16_valid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.s, vl32
; CHECK-NEXT: st1h { z0.s }, p0, [x0]
; CHECK-NEXT: ret
%a.bc = bitcast <vscale x 4 x i32> %a to <vscale x 8 x i16>
%uzp = call <vscale x 8 x i16> @llvm.aarch64.sve.uzp1.nxv8i16(<vscale x 8 x i16> %a.bc, <vscale x 8 x i16> %a.bc)
%mask = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 10)
call void @llvm.masked.store.nxv8i16(<vscale x 8 x i16> %uzp, ptr %b, i32 2, <vscale x 8 x i1> %mask)
ret void
}
define void @uzp1_i16_invalid(<vscale x 4 x i32> %a, ptr %b) #0 {
; CHECK-LABEL: uzp1_i16_invalid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.h, vl64
; CHECK-NEXT: uzp1 z0.h, z0.h, z0.h
; CHECK-NEXT: st1h { z0.h }, p0, [x0]
; CHECK-NEXT: ret
%a.bc = bitcast <vscale x 4 x i32> %a to <vscale x 8 x i16>
%uzp = call <vscale x 8 x i16> @llvm.aarch64.sve.uzp1.nxv8i16(<vscale x 8 x i16> %a.bc, <vscale x 8 x i16> %a.bc)
%mask = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 11)
call void @llvm.masked.store.nxv8i16(<vscale x 8 x i16> %uzp, ptr %b, i32 2, <vscale x 8 x i1> %mask)
ret void
}
define void @uzp1_i32_valid(<vscale x 2 x i64> %a, ptr %b) #0 {
; CHECK-LABEL: uzp1_i32_valid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.d, vl16
; CHECK-NEXT: st1w { z0.d }, p0, [x0]
; CHECK-NEXT: ret
%a.bc = bitcast <vscale x 2 x i64> %a to <vscale x 4 x i32>
%uzp = call <vscale x 4 x i32> @llvm.aarch64.sve.uzp1.nxv4i32(<vscale x 4 x i32> %a.bc, <vscale x 4 x i32> %a.bc)
%mask = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 9)
call void @llvm.masked.store.nxv4i32(<vscale x 4 x i32> %uzp, ptr %b, i32 2, <vscale x 4 x i1> %mask)
ret void
}
define void @uzp1_i32_invalid(<vscale x 2 x i64> %a, ptr %b) #0 {
; CHECK-LABEL: uzp1_i32_invalid:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.s, vl32
; CHECK-NEXT: uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT: st1w { z0.s }, p0, [x0]
; CHECK-NEXT: ret
%a.bc = bitcast <vscale x 2 x i64> %a to <vscale x 4 x i32>
%uzp = call <vscale x 4 x i32> @llvm.aarch64.sve.uzp1.nxv4i32(<vscale x 4 x i32> %a.bc, <vscale x 4 x i32> %a.bc)
%mask = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 10)
call void @llvm.masked.store.nxv4i32(<vscale x 4 x i32> %uzp, ptr %b, i32 2, <vscale x 4 x i1> %mask)
ret void
}
define void @uzp1_invalid_all(<vscale x 2 x i64> %a, ptr %b) #0 {
; CHECK-LABEL: uzp1_invalid_all:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.s
; CHECK-NEXT: uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT: st1w { z0.s }, p0, [x0]
; CHECK-NEXT: ret
%a.bc = bitcast <vscale x 2 x i64> %a to <vscale x 4 x i32>
%uzp = call <vscale x 4 x i32> @llvm.aarch64.sve.uzp1.nxv4i32(<vscale x 4 x i32> %a.bc, <vscale x 4 x i32> %a.bc)
%mask = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
call void @llvm.masked.store.nxv4i32(<vscale x 4 x i32> %uzp, ptr %b, i32 2, <vscale x 4 x i1> %mask)
ret void
}
declare <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 %pattern)
declare <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 %pattern)
declare <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 %pattern)
declare <vscale x 8 x i16> @llvm.aarch64.sve.uunpklo.nxv8i16(<vscale x 16 x i8>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.uunpklo.nxv4i32(<vscale x 8 x i16>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.uunpklo.nxv2i64(<vscale x 4 x i32>)
declare <vscale x 16 x i8> @llvm.aarch64.sve.uzp1.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.uzp1.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.uzp1.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 16 x i8> @llvm.masked.load.nxv16i8(<vscale x 16 x i8>*, i32, <vscale x 16 x i1>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.masked.load.nxv8i16(<vscale x 8 x i16>*, i32, <vscale x 8 x i1>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.masked.load.nxv4i32(<vscale x 4 x i32>*, i32, <vscale x 4 x i1>, <vscale x 4 x i32>)
declare void @llvm.masked.store.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>*, i32, <vscale x 16 x i1>)
declare void @llvm.masked.store.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>*, i32, <vscale x 8 x i1>)
declare void @llvm.masked.store.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>*, i32, <vscale x 4 x i1>)
attributes #0 = { "target-features"="+sve" vscale_range(8,0) }
|
