sve-pred-non-temporal-ldst-addressing-mode-reg-imm.ll
11.3 KB
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
; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve --asm-verbose=false < %s 2>%t | FileCheck %s
; RUN: FileCheck --check-prefix=WARN --allow-empty %s <%t
; If this check fails please read test/CodeGen/AArch64/README for instructions on how to resolve it.
; WARN-NOT: warning
; Range checks: for all the instruction tested in this file, the
; immediate must be within the range [-8, 7] (4-bit immediate). Out of
; range values are tested only in one case (following). Valid values
; are tested all through the rest of the file.
define void @imm_out_of_range(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: imm_out_of_range:
; CHECK-NEXT: rdvl x8, #8
; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: ldnt1d { z[[DATA:[0-9]+]].d }, p0/z, [x{{[0-9]+}}]
; CHECK-NEXT: rdvl x8, #-9
; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: stnt1d { z[[DATA]].d }, p0, [x{{[0-9]+}}]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 8
%base_load_bc = bitcast <vscale x 2 x i64>* %base_load to i64*
%data = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnt1.nxv2i64(<vscale x 2 x i1> %mask,
i64* %base_load_bc)
%base_store = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64> * %base, i64 -9
%base_store_bc = bitcast <vscale x 2 x i64>* %base_store to i64*
call void @llvm.aarch64.sve.stnt1.nxv2i64(<vscale x 2 x i64> %data,
<vscale x 2 x i1> %mask,
i64* %base_store_bc)
ret void
}
; 2-lane non-temporal load/stores
define void @test_masked_ldst_sv2i64(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2i64:
; CHECK-NEXT: ldnt1d { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: stnt1d { z[[DATA]].d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 -8
%base_load_bc = bitcast <vscale x 2 x i64>* %base_load to i64*
%data = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnt1.nxv2i64(<vscale x 2 x i1> %mask,
i64* %base_load_bc)
%base_store = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64> * %base, i64 -7
%base_store_bc = bitcast <vscale x 2 x i64>* %base_store to i64*
call void @llvm.aarch64.sve.stnt1.nxv2i64(<vscale x 2 x i64> %data,
<vscale x 2 x i1> %mask,
i64* %base_store_bc)
ret void
}
define void @test_masked_ldst_sv2f64(<vscale x 2 x double> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2f64:
; CHECK-NEXT: ldnt1d { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-6, mul vl]
; CHECK-NEXT: stnt1d { z[[DATA]].d }, p0, [x0, #-5, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %base, i64 -6
%base_load_bc = bitcast <vscale x 2 x double>* %base_load to double*
%data = call <vscale x 2 x double> @llvm.aarch64.sve.ldnt1.nxv2f64(<vscale x 2 x i1> %mask,
double* %base_load_bc)
%base_store = getelementptr <vscale x 2 x double>, <vscale x 2 x double> * %base, i64 -5
%base_store_bc = bitcast <vscale x 2 x double>* %base_store to double*
call void @llvm.aarch64.sve.stnt1.nxv2f64(<vscale x 2 x double> %data,
<vscale x 2 x i1> %mask,
double* %base_store_bc)
ret void
}
; 4-lane non-temporal load/stores.
define void @test_masked_ldst_sv4i32(<vscale x 4 x i32> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4i32:
; CHECK-NEXT: ldnt1w { z[[DATA:[0-9]+]].s }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: stnt1w { z[[DATA]].s }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %base, i64 6
%base_load_bc = bitcast <vscale x 4 x i32>* %base_load to i32*
%data = call <vscale x 4 x i32> @llvm.aarch64.sve.ldnt1.nxv4i32(<vscale x 4 x i1> %mask,
i32* %base_load_bc)
%base_store = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32> * %base, i64 7
%base_store_bc = bitcast <vscale x 4 x i32>* %base_store to i32*
call void @llvm.aarch64.sve.stnt1.nxv4i32(<vscale x 4 x i32> %data,
<vscale x 4 x i1> %mask,
i32* %base_store_bc)
ret void
}
define void @test_masked_ldst_sv4f32(<vscale x 4 x float> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4f32:
; CHECK-NEXT: ldnt1w { z[[DATA:[0-9]+]].s }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: stnt1w { z[[DATA]].s }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %base, i64 -1
%base_load_bc = bitcast <vscale x 4 x float>* %base_load to float*
%data = call <vscale x 4 x float> @llvm.aarch64.sve.ldnt1.nxv4f32(<vscale x 4 x i1> %mask,
float* %base_load_bc)
%base_store = getelementptr <vscale x 4 x float>, <vscale x 4 x float> * %base, i64 2
%base_store_bc = bitcast <vscale x 4 x float>* %base_store to float*
call void @llvm.aarch64.sve.stnt1.nxv4f32(<vscale x 4 x float> %data,
<vscale x 4 x i1> %mask,
float* %base_store_bc)
ret void
}
; 8-lane non-temporal load/stores.
define void @test_masked_ldst_sv8i16(<vscale x 8 x i16> * %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv8i16:
; CHECK-NEXT: ldnt1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: stnt1h { z[[DATA]].h }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %base, i64 6
%base_load_bc = bitcast <vscale x 8 x i16>* %base_load to i16*
%data = call <vscale x 8 x i16> @llvm.aarch64.sve.ldnt1.nxv8i16(<vscale x 8 x i1> %mask,
i16* %base_load_bc)
%base_store = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16> * %base, i64 7
%base_store_bc = bitcast <vscale x 8 x i16>* %base_store to i16*
call void @llvm.aarch64.sve.stnt1.nxv8i16(<vscale x 8 x i16> %data,
<vscale x 8 x i1> %mask,
i16* %base_store_bc)
ret void
}
define void @test_masked_ldst_sv8f16(<vscale x 8 x half> * %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv8f16:
; CHECK-NEXT: ldnt1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: stnt1h { z[[DATA]].h }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %base, i64 -1
%base_load_bc = bitcast <vscale x 8 x half>* %base_load to half*
%data = call <vscale x 8 x half> @llvm.aarch64.sve.ldnt1.nxv8f16(<vscale x 8 x i1> %mask,
half* %base_load_bc)
%base_store = getelementptr <vscale x 8 x half>, <vscale x 8 x half> * %base, i64 2
%base_store_bc = bitcast <vscale x 8 x half>* %base_store to half*
call void @llvm.aarch64.sve.stnt1.nxv8f16(<vscale x 8 x half> %data,
<vscale x 8 x i1> %mask,
half* %base_store_bc)
ret void
}
define void @test_masked_ldst_sv8bf16(<vscale x 8 x bfloat> * %base, <vscale x 8 x i1> %mask) nounwind #0 {
; CHECK-LABEL: test_masked_ldst_sv8bf16:
; CHECK-NEXT: ldnt1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: stnt1h { z[[DATA]].h }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat>* %base, i64 -1
%base_load_bc = bitcast <vscale x 8 x bfloat>* %base_load to bfloat*
%data = call <vscale x 8 x bfloat> @llvm.aarch64.sve.ldnt1.nxv8bf16(<vscale x 8 x i1> %mask,
bfloat* %base_load_bc)
%base_store = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat> * %base, i64 2
%base_store_bc = bitcast <vscale x 8 x bfloat>* %base_store to bfloat*
call void @llvm.aarch64.sve.stnt1.nxv8bf16(<vscale x 8 x bfloat> %data,
<vscale x 8 x i1> %mask,
bfloat* %base_store_bc)
ret void
}
; 16-lane non-temporal load/stores.
define void @test_masked_ldst_sv16i8(<vscale x 16 x i8> * %base, <vscale x 16 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv16i8:
; CHECK-NEXT: ldnt1b { z[[DATA:[0-9]+]].b }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: stnt1b { z[[DATA]].b }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base, i64 6
%base_load_bc = bitcast <vscale x 16 x i8>* %base_load to i8*
%data = call <vscale x 16 x i8> @llvm.aarch64.sve.ldnt1.nxv16i8(<vscale x 16 x i1> %mask,
i8* %base_load_bc)
%base_store = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8> * %base, i64 7
%base_store_bc = bitcast <vscale x 16 x i8>* %base_store to i8*
call void @llvm.aarch64.sve.stnt1.nxv16i8(<vscale x 16 x i8> %data,
<vscale x 16 x i1> %mask,
i8* %base_store_bc)
ret void
}
; 2-element non-temporal loads.
declare <vscale x 2 x i64> @llvm.aarch64.sve.ldnt1.nxv2i64(<vscale x 2 x i1>, i64*)
declare <vscale x 2 x double> @llvm.aarch64.sve.ldnt1.nxv2f64(<vscale x 2 x i1>, double*)
; 4-element non-temporal loads.
declare <vscale x 4 x i32> @llvm.aarch64.sve.ldnt1.nxv4i32(<vscale x 4 x i1>, i32*)
declare <vscale x 4 x float> @llvm.aarch64.sve.ldnt1.nxv4f32(<vscale x 4 x i1>, float*)
; 8-element non-temporal loads.
declare <vscale x 8 x i16> @llvm.aarch64.sve.ldnt1.nxv8i16(<vscale x 8 x i1>, i16*)
declare <vscale x 8 x half> @llvm.aarch64.sve.ldnt1.nxv8f16(<vscale x 8 x i1>, half*)
declare <vscale x 8 x bfloat> @llvm.aarch64.sve.ldnt1.nxv8bf16(<vscale x 8 x i1>, bfloat*)
; 16-element non-temporal loads.
declare <vscale x 16 x i8> @llvm.aarch64.sve.ldnt1.nxv16i8(<vscale x 16 x i1>, i8*)
; 2-element non-temporal stores.
declare void @llvm.aarch64.sve.stnt1.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i1>, i64*)
declare void @llvm.aarch64.sve.stnt1.nxv2f64(<vscale x 2 x double>, <vscale x 2 x i1>, double*)
; 4-element non-temporal stores.
declare void @llvm.aarch64.sve.stnt1.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i1>, i32*)
declare void @llvm.aarch64.sve.stnt1.nxv4f32(<vscale x 4 x float>, <vscale x 4 x i1>, float*)
; 8-element non-temporal stores.
declare void @llvm.aarch64.sve.stnt1.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i1>, i16*)
declare void @llvm.aarch64.sve.stnt1.nxv8f16(<vscale x 8 x half>, <vscale x 8 x i1>, half*)
declare void @llvm.aarch64.sve.stnt1.nxv8bf16(<vscale x 8 x bfloat>, <vscale x 8 x i1>, bfloat*)
; 16-element non-temporal stores.
declare void @llvm.aarch64.sve.stnt1.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i1>, i8*)
; +bf16 is required for the bfloat version.
attributes #0 = { "target-features"="+sve,+bf16" }