acle_sve_ld4.c
12.1 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
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
#include <arm_sve.h>
#ifdef SVE_OVERLOADED_FORMS
// A simple used,unused... macro, long enough to represent any SVE builtin.
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
#else
#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
#endif
svint8x4_t test_svld4_s8(svbool_t pg, const int8_t *base)
{
// CHECK-LABEL: test_svld4_s8
// CHECK: %[[LOAD:.*]] = call <vscale x 64 x i8> @llvm.aarch64.sve.ld4.nxv64i8.nxv16i1(<vscale x 16 x i1> %pg, i8* %base)
// CHECK-NEXT: ret <vscale x 64 x i8> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_s8,,)(pg, base);
}
svint16x4_t test_svld4_s16(svbool_t pg, const int16_t *base)
{
// CHECK-LABEL: test_svld4_s16
// CHECK: %[[PG:.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 32 x i16> @llvm.aarch64.sve.ld4.nxv32i16.nxv8i1(<vscale x 8 x i1> %[[PG]], i16* %base)
// CHECK-NEXT: ret <vscale x 32 x i16> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_s16,,)(pg, base);
}
svint32x4_t test_svld4_s32(svbool_t pg, const int32_t *base)
{
// CHECK-LABEL: test_svld4_s32
// CHECK: %[[PG:.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 16 x i32> @llvm.aarch64.sve.ld4.nxv16i32.nxv4i1(<vscale x 4 x i1> %[[PG]], i32* %base)
// CHECK-NEXT: ret <vscale x 16 x i32> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_s32,,)(pg, base);
}
svint64x4_t test_svld4_s64(svbool_t pg, const int64_t *base)
{
// CHECK-LABEL: test_svld4_s64
// CHECK: %[[PG:.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 8 x i64> @llvm.aarch64.sve.ld4.nxv8i64.nxv2i1(<vscale x 2 x i1> %[[PG]], i64* %base)
// CHECK-NEXT: ret <vscale x 8 x i64> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_s64,,)(pg, base);
}
svuint8x4_t test_svld4_u8(svbool_t pg, const uint8_t *base)
{
// CHECK-LABEL: test_svld4_u8
// CHECK: %[[LOAD:.*]] = call <vscale x 64 x i8> @llvm.aarch64.sve.ld4.nxv64i8.nxv16i1(<vscale x 16 x i1> %pg, i8* %base)
// CHECK-NEXT: ret <vscale x 64 x i8> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_u8,,)(pg, base);
}
svuint16x4_t test_svld4_u16(svbool_t pg, const uint16_t *base)
{
// CHECK-LABEL: test_svld4_u16
// CHECK: %[[PG:.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 32 x i16> @llvm.aarch64.sve.ld4.nxv32i16.nxv8i1(<vscale x 8 x i1> %[[PG]], i16* %base)
// CHECK-NEXT: ret <vscale x 32 x i16> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_u16,,)(pg, base);
}
svuint32x4_t test_svld4_u32(svbool_t pg, const uint32_t *base)
{
// CHECK-LABEL: test_svld4_u32
// CHECK: %[[PG:.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 16 x i32> @llvm.aarch64.sve.ld4.nxv16i32.nxv4i1(<vscale x 4 x i1> %[[PG]], i32* %base)
// CHECK-NEXT: ret <vscale x 16 x i32> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_u32,,)(pg, base);
}
svuint64x4_t test_svld4_u64(svbool_t pg, const uint64_t *base)
{
// CHECK-LABEL: test_svld4_u64
// CHECK: %[[PG:.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 8 x i64> @llvm.aarch64.sve.ld4.nxv8i64.nxv2i1(<vscale x 2 x i1> %[[PG]], i64* %base)
// CHECK-NEXT: ret <vscale x 8 x i64> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_u64,,)(pg, base);
}
svfloat16x4_t test_svld4_f16(svbool_t pg, const float16_t *base)
{
// CHECK-LABEL: test_svld4_f16
// CHECK: %[[PG:.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 32 x half> @llvm.aarch64.sve.ld4.nxv32f16.nxv8i1(<vscale x 8 x i1> %[[PG]], half* %base)
// CHECK-NEXT: ret <vscale x 32 x half> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_f16,,)(pg, base);
}
svfloat32x4_t test_svld4_f32(svbool_t pg, const float32_t *base)
{
// CHECK-LABEL: test_svld4_f32
// CHECK: %[[PG:.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 16 x float> @llvm.aarch64.sve.ld4.nxv16f32.nxv4i1(<vscale x 4 x i1> %[[PG]], float* %base)
// CHECK-NEXT: ret <vscale x 16 x float> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_f32,,)(pg, base);
}
svfloat64x4_t test_svld4_f64(svbool_t pg, const float64_t *base)
{
// CHECK-LABEL: test_svld4_f64
// CHECK: %[[PG:.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> %pg)
// CHECK: %[[LOAD:.*]] = call <vscale x 8 x double> @llvm.aarch64.sve.ld4.nxv8f64.nxv2i1(<vscale x 2 x i1> %[[PG]], double* %base)
// CHECK-NEXT: ret <vscale x 8 x double> %[[LOAD]]
return SVE_ACLE_FUNC(svld4,_f64,,)(pg, base);
}
svint8x4_t test_svld4_vnum_s8(svbool_t pg, const int8_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_s8
// CHECK: %[[BASE:.*]] = bitcast i8* %base to <vscale x 16 x i8>*
// CHECK: %[[GEP:.*]] = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 64 x i8> @llvm.aarch64.sve.ld4.nxv64i8.nxv16i1(<vscale x 16 x i1> %pg, i8* %[[GEP]])
// CHECK-NEXT: ret <vscale x 64 x i8> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_s8,,)(pg, base, vnum);
}
svint16x4_t test_svld4_vnum_s16(svbool_t pg, const int16_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_s16
// CHECK-DAG: %[[PG:.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast i16* %base to <vscale x 8 x i16>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 32 x i16> @llvm.aarch64.sve.ld4.nxv32i16.nxv8i1(<vscale x 8 x i1> %[[PG]], i16* %[[GEP]])
// CHECK-NEXT: ret <vscale x 32 x i16> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_s16,,)(pg, base, vnum);
}
svint32x4_t test_svld4_vnum_s32(svbool_t pg, const int32_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_s32
// CHECK-DAG: %[[PG:.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast i32* %base to <vscale x 4 x i32>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 16 x i32> @llvm.aarch64.sve.ld4.nxv16i32.nxv4i1(<vscale x 4 x i1> %[[PG]], i32* %[[GEP]])
// CHECK-NEXT: ret <vscale x 16 x i32> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_s32,,)(pg, base, vnum);
}
svint64x4_t test_svld4_vnum_s64(svbool_t pg, const int64_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_s64
// CHECK-DAG: %[[PG:.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast i64* %base to <vscale x 2 x i64>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 8 x i64> @llvm.aarch64.sve.ld4.nxv8i64.nxv2i1(<vscale x 2 x i1> %[[PG]], i64* %[[GEP]])
// CHECK-NEXT: ret <vscale x 8 x i64> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_s64,,)(pg, base, vnum);
}
svuint8x4_t test_svld4_vnum_u8(svbool_t pg, const uint8_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_u8
// CHECK: %[[BASE:.*]] = bitcast i8* %base to <vscale x 16 x i8>*
// CHECK: %[[GEP:.*]] = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 64 x i8> @llvm.aarch64.sve.ld4.nxv64i8.nxv16i1(<vscale x 16 x i1> %pg, i8* %[[GEP]])
// CHECK-NEXT: ret <vscale x 64 x i8> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_u8,,)(pg, base, vnum);
}
svuint16x4_t test_svld4_vnum_u16(svbool_t pg, const uint16_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_u16
// CHECK-DAG: %[[PG:.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast i16* %base to <vscale x 8 x i16>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 32 x i16> @llvm.aarch64.sve.ld4.nxv32i16.nxv8i1(<vscale x 8 x i1> %[[PG]], i16* %[[GEP]])
// CHECK-NEXT: ret <vscale x 32 x i16> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_u16,,)(pg, base, vnum);
}
svuint32x4_t test_svld4_vnum_u32(svbool_t pg, const uint32_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_u32
// CHECK-DAG: %[[PG:.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast i32* %base to <vscale x 4 x i32>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 16 x i32> @llvm.aarch64.sve.ld4.nxv16i32.nxv4i1(<vscale x 4 x i1> %[[PG]], i32* %[[GEP]])
// CHECK-NEXT: ret <vscale x 16 x i32> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_u32,,)(pg, base, vnum);
}
svuint64x4_t test_svld4_vnum_u64(svbool_t pg, const uint64_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_u64
// CHECK-DAG: %[[PG:.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast i64* %base to <vscale x 2 x i64>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 8 x i64> @llvm.aarch64.sve.ld4.nxv8i64.nxv2i1(<vscale x 2 x i1> %[[PG]], i64* %[[GEP]])
// CHECK-NEXT: ret <vscale x 8 x i64> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_u64,,)(pg, base, vnum);
}
svfloat16x4_t test_svld4_vnum_f16(svbool_t pg, const float16_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_f16
// CHECK-DAG: %[[PG:.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast half* %base to <vscale x 8 x half>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 32 x half> @llvm.aarch64.sve.ld4.nxv32f16.nxv8i1(<vscale x 8 x i1> %[[PG]], half* %[[GEP]])
// CHECK-NEXT: ret <vscale x 32 x half> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_f16,,)(pg, base, vnum);
}
svfloat32x4_t test_svld4_vnum_f32(svbool_t pg, const float32_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_f32
// CHECK-DAG: %[[PG:.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast float* %base to <vscale x 4 x float>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 16 x float> @llvm.aarch64.sve.ld4.nxv16f32.nxv4i1(<vscale x 4 x i1> %[[PG]], float* %[[GEP]])
// CHECK-NEXT: ret <vscale x 16 x float> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_f32,,)(pg, base, vnum);
}
svfloat64x4_t test_svld4_vnum_f64(svbool_t pg, const float64_t *base, int64_t vnum)
{
// CHECK-LABEL: test_svld4_vnum_f64
// CHECK-DAG: %[[PG:.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> %pg)
// CHECK-DAG: %[[BASE:.*]] = bitcast double* %base to <vscale x 2 x double>*
// CHECK-DAG: %[[GEP:.*]] = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %[[BASE]], i64 %vnum, i64 0
// CHECK: %[[LOAD:.*]] = call <vscale x 8 x double> @llvm.aarch64.sve.ld4.nxv8f64.nxv2i1(<vscale x 2 x i1> %[[PG]], double* %[[GEP]])
// CHECK-NEXT: ret <vscale x 8 x double> %[[LOAD]]
return SVE_ACLE_FUNC(svld4_vnum,_f64,,)(pg, base, vnum);
}