vshlc.c
9.86 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
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// RUN: %clang_cc1 -triple thumbv8.1m.main-none-none-eabi -target-feature +mve -mfloat-abi hard -fallow-half-arguments-and-returns -O0 -disable-O0-optnone -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
// RUN: %clang_cc1 -triple thumbv8.1m.main-none-none-eabi -target-feature +mve -mfloat-abi hard -fallow-half-arguments-and-returns -O0 -disable-O0-optnone -DPOLYMORPHIC -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
#include <arm_mve.h>
// CHECK-LABEL: @test_vshlcq_s8(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = call { i32, <16 x i8> } @llvm.arm.mve.vshlc.v16i8(<16 x i8> [[A:%.*]], i32 [[TMP0]], i32 18)
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { i32, <16 x i8> } [[TMP1]], 0
// CHECK-NEXT: store i32 [[TMP2]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i32, <16 x i8> } [[TMP1]], 1
// CHECK-NEXT: ret <16 x i8> [[TMP3]]
//
int8x16_t test_vshlcq_s8(int8x16_t a, uint32_t *b) {
#ifdef POLYMORPHIC
return vshlcq(a, b, 18);
#else /* POLYMORPHIC */
return vshlcq_s8(a, b, 18);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = call { i32, <8 x i16> } @llvm.arm.mve.vshlc.v8i16(<8 x i16> [[A:%.*]], i32 [[TMP0]], i32 16)
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { i32, <8 x i16> } [[TMP1]], 0
// CHECK-NEXT: store i32 [[TMP2]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i32, <8 x i16> } [[TMP1]], 1
// CHECK-NEXT: ret <8 x i16> [[TMP3]]
//
int16x8_t test_vshlcq_s16(int16x8_t a, uint32_t *b) {
#ifdef POLYMORPHIC
return vshlcq(a, b, 16);
#else /* POLYMORPHIC */
return vshlcq_s16(a, b, 16);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = call { i32, <4 x i32> } @llvm.arm.mve.vshlc.v4i32(<4 x i32> [[A:%.*]], i32 [[TMP0]], i32 4)
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { i32, <4 x i32> } [[TMP1]], 0
// CHECK-NEXT: store i32 [[TMP2]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i32, <4 x i32> } [[TMP1]], 1
// CHECK-NEXT: ret <4 x i32> [[TMP3]]
//
int32x4_t test_vshlcq_s32(int32x4_t a, uint32_t *b) {
#ifdef POLYMORPHIC
return vshlcq(a, b, 4);
#else /* POLYMORPHIC */
return vshlcq_s32(a, b, 4);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_u8(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = call { i32, <16 x i8> } @llvm.arm.mve.vshlc.v16i8(<16 x i8> [[A:%.*]], i32 [[TMP0]], i32 17)
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { i32, <16 x i8> } [[TMP1]], 0
// CHECK-NEXT: store i32 [[TMP2]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i32, <16 x i8> } [[TMP1]], 1
// CHECK-NEXT: ret <16 x i8> [[TMP3]]
//
uint8x16_t test_vshlcq_u8(uint8x16_t a, uint32_t *b) {
#ifdef POLYMORPHIC
return vshlcq(a, b, 17);
#else /* POLYMORPHIC */
return vshlcq_u8(a, b, 17);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = call { i32, <8 x i16> } @llvm.arm.mve.vshlc.v8i16(<8 x i16> [[A:%.*]], i32 [[TMP0]], i32 17)
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { i32, <8 x i16> } [[TMP1]], 0
// CHECK-NEXT: store i32 [[TMP2]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i32, <8 x i16> } [[TMP1]], 1
// CHECK-NEXT: ret <8 x i16> [[TMP3]]
//
uint16x8_t test_vshlcq_u16(uint16x8_t a, uint32_t *b) {
#ifdef POLYMORPHIC
return vshlcq(a, b, 17);
#else /* POLYMORPHIC */
return vshlcq_u16(a, b, 17);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = call { i32, <4 x i32> } @llvm.arm.mve.vshlc.v4i32(<4 x i32> [[A:%.*]], i32 [[TMP0]], i32 20)
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { i32, <4 x i32> } [[TMP1]], 0
// CHECK-NEXT: store i32 [[TMP2]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i32, <4 x i32> } [[TMP1]], 1
// CHECK-NEXT: ret <4 x i32> [[TMP3]]
//
uint32x4_t test_vshlcq_u32(uint32x4_t a, uint32_t *b) {
#ifdef POLYMORPHIC
return vshlcq(a, b, 20);
#else /* POLYMORPHIC */
return vshlcq_u32(a, b, 20);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_m_s8(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT: [[TMP2:%.*]] = call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call { i32, <16 x i8> } @llvm.arm.mve.vshlc.predicated.v16i8.v16i1(<16 x i8> [[A:%.*]], i32 [[TMP0]], i32 29, <16 x i1> [[TMP2]])
// CHECK-NEXT: [[TMP4:%.*]] = extractvalue { i32, <16 x i8> } [[TMP3]], 0
// CHECK-NEXT: store i32 [[TMP4]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, <16 x i8> } [[TMP3]], 1
// CHECK-NEXT: ret <16 x i8> [[TMP5]]
//
int8x16_t test_vshlcq_m_s8(int8x16_t a, uint32_t *b, mve_pred16_t p) {
#ifdef POLYMORPHIC
return vshlcq_m(a, b, 29, p);
#else /* POLYMORPHIC */
return vshlcq_m_s8(a, b, 29, p);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_m_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT: [[TMP2:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call { i32, <8 x i16> } @llvm.arm.mve.vshlc.predicated.v8i16.v8i1(<8 x i16> [[A:%.*]], i32 [[TMP0]], i32 17, <8 x i1> [[TMP2]])
// CHECK-NEXT: [[TMP4:%.*]] = extractvalue { i32, <8 x i16> } [[TMP3]], 0
// CHECK-NEXT: store i32 [[TMP4]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, <8 x i16> } [[TMP3]], 1
// CHECK-NEXT: ret <8 x i16> [[TMP5]]
//
int16x8_t test_vshlcq_m_s16(int16x8_t a, uint32_t *b, mve_pred16_t p) {
#ifdef POLYMORPHIC
return vshlcq_m(a, b, 17, p);
#else /* POLYMORPHIC */
return vshlcq_m_s16(a, b, 17, p);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_m_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT: [[TMP2:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call { i32, <4 x i32> } @llvm.arm.mve.vshlc.predicated.v4i32.v4i1(<4 x i32> [[A:%.*]], i32 [[TMP0]], i32 9, <4 x i1> [[TMP2]])
// CHECK-NEXT: [[TMP4:%.*]] = extractvalue { i32, <4 x i32> } [[TMP3]], 0
// CHECK-NEXT: store i32 [[TMP4]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, <4 x i32> } [[TMP3]], 1
// CHECK-NEXT: ret <4 x i32> [[TMP5]]
//
int32x4_t test_vshlcq_m_s32(int32x4_t a, uint32_t *b, mve_pred16_t p) {
#ifdef POLYMORPHIC
return vshlcq_m(a, b, 9, p);
#else /* POLYMORPHIC */
return vshlcq_m_s32(a, b, 9, p);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_m_u8(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT: [[TMP2:%.*]] = call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call { i32, <16 x i8> } @llvm.arm.mve.vshlc.predicated.v16i8.v16i1(<16 x i8> [[A:%.*]], i32 [[TMP0]], i32 21, <16 x i1> [[TMP2]])
// CHECK-NEXT: [[TMP4:%.*]] = extractvalue { i32, <16 x i8> } [[TMP3]], 0
// CHECK-NEXT: store i32 [[TMP4]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, <16 x i8> } [[TMP3]], 1
// CHECK-NEXT: ret <16 x i8> [[TMP5]]
//
uint8x16_t test_vshlcq_m_u8(uint8x16_t a, uint32_t *b, mve_pred16_t p) {
#ifdef POLYMORPHIC
return vshlcq_m(a, b, 21, p);
#else /* POLYMORPHIC */
return vshlcq_m_u8(a, b, 21, p);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_m_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT: [[TMP2:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call { i32, <8 x i16> } @llvm.arm.mve.vshlc.predicated.v8i16.v8i1(<8 x i16> [[A:%.*]], i32 [[TMP0]], i32 24, <8 x i1> [[TMP2]])
// CHECK-NEXT: [[TMP4:%.*]] = extractvalue { i32, <8 x i16> } [[TMP3]], 0
// CHECK-NEXT: store i32 [[TMP4]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, <8 x i16> } [[TMP3]], 1
// CHECK-NEXT: ret <8 x i16> [[TMP5]]
//
uint16x8_t test_vshlcq_m_u16(uint16x8_t a, uint32_t *b, mve_pred16_t p) {
#ifdef POLYMORPHIC
return vshlcq_m(a, b, 24, p);
#else /* POLYMORPHIC */
return vshlcq_m_u16(a, b, 24, p);
#endif /* POLYMORPHIC */
}
// CHECK-LABEL: @test_vshlcq_m_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, i32* [[B:%.*]], align 4
// CHECK-NEXT: [[TMP1:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT: [[TMP2:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call { i32, <4 x i32> } @llvm.arm.mve.vshlc.predicated.v4i32.v4i1(<4 x i32> [[A:%.*]], i32 [[TMP0]], i32 26, <4 x i1> [[TMP2]])
// CHECK-NEXT: [[TMP4:%.*]] = extractvalue { i32, <4 x i32> } [[TMP3]], 0
// CHECK-NEXT: store i32 [[TMP4]], i32* [[B]], align 4
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, <4 x i32> } [[TMP3]], 1
// CHECK-NEXT: ret <4 x i32> [[TMP5]]
//
uint32x4_t test_vshlcq_m_u32(uint32x4_t a, uint32_t *b, mve_pred16_t p) {
#ifdef POLYMORPHIC
return vshlcq_m(a, b, 26, p);
#else /* POLYMORPHIC */
return vshlcq_m_u32(a, b, 26, p);
#endif /* POLYMORPHIC */
}