Debian Codegraph @

clang/test/CodeGen/AArch64/sve2p1-intrinsics/acle_sve2p1_store_struct.c

raw
1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
2 // REQUIRES: aarch64-registered-target
3 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
4 // RUN: %clang_cc1 -triple aarch64 -target-feature +sme -target-feature +sme2 -target-feature +sme2p1 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
5 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sme -target-feature +sve2p1 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
6 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sme -target-feature +sme2p1 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
7 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
8 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
9 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
10 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
11 #include <arm_sve.h>
12
13 #ifdef SVE_OVERLOADED_FORMS
14 // A simple used,unused... macro, long enough to represent any SVE builtin.
15 #define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
16 #else
17 #define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
18 #endif
19
20 #if defined(__ARM_FEATURE_SME) && defined(__ARM_FEATURE_SVE)
21 #define ATTR __arm_streaming_compatible
22 #elif defined(__ARM_FEATURE_SME)
23 #define ATTR __arm_streaming
24 #else
25 #define ATTR
26 #endif
27
28 // CHECK-LABEL: @test_svst2q_u8(
29 // CHECK-NEXT: entry:
30 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
31 // CHECK-NEXT: ret void
32 //
33 // CPP-CHECK-LABEL: @_Z14test_svst2q_u8u10__SVBool_tPKh11svuint8x2_t(
34 // CPP-CHECK-NEXT: entry:
35 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
36 // CPP-CHECK-NEXT: ret void
37 //
38 void test_svst2q_u8(svbool_t pg, const uint8_t *base, svuint8x2_t zt) ATTR
39 {
40 SVE_ACLE_FUNC(svst2q,,_u8,)(pg, base, zt);
41 }
42
43 // CHECK-LABEL: @test_svst2q_s8(
44 // CHECK-NEXT: entry:
45 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
46 // CHECK-NEXT: ret void
47 //
48 // CPP-CHECK-LABEL: @_Z14test_svst2q_s8u10__SVBool_tPKa10svint8x2_t(
49 // CPP-CHECK-NEXT: entry:
50 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
51 // CPP-CHECK-NEXT: ret void
52 //
53 void test_svst2q_s8(svbool_t pg, const int8_t *base, svint8x2_t zt) ATTR
54 {
55 SVE_ACLE_FUNC(svst2q,,_s8,)(pg, base, zt);
56 }
57 // CHECK-LABEL: @test_svst2q_u16(
58 // CHECK-NEXT: entry:
59 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
60 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
61 // CHECK-NEXT: ret void
62 //
63 // CPP-CHECK-LABEL: @_Z15test_svst2q_u16u10__SVBool_tPKt12svuint16x2_t(
64 // CPP-CHECK-NEXT: entry:
65 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
66 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
67 // CPP-CHECK-NEXT: ret void
68 //
69 void test_svst2q_u16(svbool_t pg, const uint16_t *base, svuint16x2_t zt) ATTR
70 {
71 SVE_ACLE_FUNC(svst2q,,_u16,)(pg, base, zt);
72 }
73
74 // CHECK-LABEL: @test_svst2q_s16(
75 // CHECK-NEXT: entry:
76 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
77 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
78 // CHECK-NEXT: ret void
79 //
80 // CPP-CHECK-LABEL: @_Z15test_svst2q_s16u10__SVBool_tPKs11svint16x2_t(
81 // CPP-CHECK-NEXT: entry:
82 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
83 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
84 // CPP-CHECK-NEXT: ret void
85 //
86 void test_svst2q_s16(svbool_t pg, const int16_t *base, svint16x2_t zt) ATTR
87 {
88 SVE_ACLE_FUNC(svst2q,,_s16,)(pg, base, zt);
89 }
90
91 // CHECK-LABEL: @test_svst2q_u32(
92 // CHECK-NEXT: entry:
93 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
94 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
95 // CHECK-NEXT: ret void
96 //
97 // CPP-CHECK-LABEL: @_Z15test_svst2q_u32u10__SVBool_tPKj12svuint32x2_t(
98 // CPP-CHECK-NEXT: entry:
99 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
100 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
101 // CPP-CHECK-NEXT: ret void
102 //
103 void test_svst2q_u32(svbool_t pg, const uint32_t *base, svuint32x2_t zt) ATTR
104 {
105 SVE_ACLE_FUNC(svst2q,,_u32,)(pg, base, zt);
106 }
107
108 // CHECK-LABEL: @test_svst2q_s32(
109 // CHECK-NEXT: entry:
110 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
111 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
112 // CHECK-NEXT: ret void
113 //
114 // CPP-CHECK-LABEL: @_Z15test_svst2q_s32u10__SVBool_tPKi11svint32x2_t(
115 // CPP-CHECK-NEXT: entry:
116 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
117 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
118 // CPP-CHECK-NEXT: ret void
119 //
120 void test_svst2q_s32(svbool_t pg, const int32_t *base, svint32x2_t zt) ATTR
121 {
122 SVE_ACLE_FUNC(svst2q,,_s32,)(pg, base, zt);
123 }
124
125 // CHECK-LABEL: @test_svst2q_u64(
126 // CHECK-NEXT: entry:
127 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
128 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
129 // CHECK-NEXT: ret void
130 //
131 // CPP-CHECK-LABEL: @_Z15test_svst2q_u64u10__SVBool_tPKm12svuint64x2_t(
132 // CPP-CHECK-NEXT: entry:
133 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
134 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
135 // CPP-CHECK-NEXT: ret void
136 //
137 void test_svst2q_u64(svbool_t pg, const uint64_t *base, svuint64x2_t zt) ATTR
138 {
139 SVE_ACLE_FUNC(svst2q,,_u64,)(pg, base, zt);
140 }
141
142 // CHECK-LABEL: @test_svst2q_s64(
143 // CHECK-NEXT: entry:
144 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
145 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
146 // CHECK-NEXT: ret void
147 //
148 // CPP-CHECK-LABEL: @_Z15test_svst2q_s64u10__SVBool_tPKl11svint64x2_t(
149 // CPP-CHECK-NEXT: entry:
150 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
151 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
152 // CPP-CHECK-NEXT: ret void
153 //
154 void test_svst2q_s64(svbool_t pg, const int64_t *base, svint64x2_t zt) ATTR
155 {
156 SVE_ACLE_FUNC(svst2q,,_s64,)(pg, base, zt);
157 }
158
159 // CHECK-LABEL: @test_svst2q_f16(
160 // CHECK-NEXT: entry:
161 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
162 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
163 // CHECK-NEXT: ret void
164 //
165 // CPP-CHECK-LABEL: @_Z15test_svst2q_f16u10__SVBool_tPKDh13svfloat16x2_t(
166 // CPP-CHECK-NEXT: entry:
167 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
168 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
169 // CPP-CHECK-NEXT: ret void
170 //
171 void test_svst2q_f16(svbool_t pg, const float16_t *base, svfloat16x2_t zt) ATTR
172 {
173 SVE_ACLE_FUNC(svst2q,,_f16,)(pg, base, zt);
174 }
175
176 // CHECK-LABEL: @test_svst2q_bf16(
177 // CHECK-NEXT: entry:
178 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
179 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
180 // CHECK-NEXT: ret void
181 //
182 // CPP-CHECK-LABEL: @_Z16test_svst2q_bf16u10__SVBool_tPKu6__bf1614svbfloat16x2_t(
183 // CPP-CHECK-NEXT: entry:
184 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
185 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
186 // CPP-CHECK-NEXT: ret void
187 //
188 void test_svst2q_bf16(svbool_t pg, const bfloat16_t *base, svbfloat16x2_t zt) ATTR
189 {
190 SVE_ACLE_FUNC(svst2q,,_bf16,)(pg, base, zt);
191 }
192
193 // CHECK-LABEL: @test_svst2q_f32(
194 // CHECK-NEXT: entry:
195 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
196 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
197 // CHECK-NEXT: ret void
198 //
199 // CPP-CHECK-LABEL: @_Z15test_svst2q_f32u10__SVBool_tPKf13svfloat32x2_t(
200 // CPP-CHECK-NEXT: entry:
201 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
202 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
203 // CPP-CHECK-NEXT: ret void
204 //
205 void test_svst2q_f32(svbool_t pg, const float32_t *base, svfloat32x2_t zt) ATTR
206 {
207 SVE_ACLE_FUNC(svst2q,,_f32,)(pg, base, zt);
208 }
209
210 // CHECK-LABEL: @test_svst2q_f64(
211 // CHECK-NEXT: entry:
212 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
213 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
214 // CHECK-NEXT: ret void
215 //
216 // CPP-CHECK-LABEL: @_Z15test_svst2q_f64u10__SVBool_tPKd13svfloat64x2_t(
217 // CPP-CHECK-NEXT: entry:
218 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
219 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
220 // CPP-CHECK-NEXT: ret void
221 //
222 void test_svst2q_f64(svbool_t pg, const float64_t *base, svfloat64x2_t zt) ATTR
223 {
224 SVE_ACLE_FUNC(svst2q,,_f64,)(pg, base, zt);
225 }
226
227 // CHECK-LABEL: @test_svst2q_mf8(
228 // CHECK-NEXT: entry:
229 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
230 // CHECK-NEXT: ret void
231 //
232 // CPP-CHECK-LABEL: @_Z15test_svst2q_mf8u10__SVBool_tPKu6__mfp813svmfloat8x2_t(
233 // CPP-CHECK-NEXT: entry:
234 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
235 // CPP-CHECK-NEXT: ret void
236 //
237 void test_svst2q_mf8(svbool_t pg, const mfloat8_t *base, svmfloat8x2_t zt) ATTR
238 {
239 SVE_ACLE_FUNC(svst2q,,_mf8,)(pg, base, zt);
240 }
241
242 // CHECK-LABEL: @test_svst2q_vnum_u8(
243 // CHECK-NEXT: entry:
244 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
245 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
246 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
247 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
248 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
249 // CHECK-NEXT: ret void
250 //
251 // CPP-CHECK-LABEL: @_Z19test_svst2q_vnum_u8u10__SVBool_tPKhl11svuint8x2_t(
252 // CPP-CHECK-NEXT: entry:
253 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
254 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
255 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
256 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
257 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
258 // CPP-CHECK-NEXT: ret void
259 //
260 void test_svst2q_vnum_u8(svbool_t pg, const uint8_t *base, int64_t vnum, svuint8x2_t zt) ATTR
261 {
262 SVE_ACLE_FUNC(svst2q_vnum_,,u8,)(pg, base, vnum, zt);
263 }
264
265 // CHECK-LABEL: @test_svst2q_vnum_s8(
266 // CHECK-NEXT: entry:
267 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
268 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
269 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
270 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
271 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
272 // CHECK-NEXT: ret void
273 //
274 // CPP-CHECK-LABEL: @_Z19test_svst2q_vnum_s8u10__SVBool_tPKal10svint8x2_t(
275 // CPP-CHECK-NEXT: entry:
276 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
277 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
278 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
279 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
280 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
281 // CPP-CHECK-NEXT: ret void
282 //
283 void test_svst2q_vnum_s8(svbool_t pg, const int8_t *base, int64_t vnum, svint8x2_t zt) ATTR
284 {
285 SVE_ACLE_FUNC(svst2q_vnum,,_s8,)(pg, base, vnum, zt);
286 }
287
288 // CHECK-LABEL: @test_svst2q_vnum_u16(
289 // CHECK-NEXT: entry:
290 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
291 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
292 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
293 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
294 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
295 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
296 // CHECK-NEXT: ret void
297 //
298 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_u16u10__SVBool_tPKtl12svuint16x2_t(
299 // CPP-CHECK-NEXT: entry:
300 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
301 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
302 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
303 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
304 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
305 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
306 // CPP-CHECK-NEXT: ret void
307 //
308 void test_svst2q_vnum_u16(svbool_t pg, const uint16_t *base, int64_t vnum, svuint16x2_t zt) ATTR
309 {
310 SVE_ACLE_FUNC(svst2q_vnum,,_u16,)(pg, base, vnum, zt);
311 }
312
313 // CHECK-LABEL: @test_svst2q_vnum_s16(
314 // CHECK-NEXT: entry:
315 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
316 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
317 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
318 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
319 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
320 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
321 // CHECK-NEXT: ret void
322 //
323 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_s16u10__SVBool_tPKsl11svint16x2_t(
324 // CPP-CHECK-NEXT: entry:
325 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
326 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
327 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
328 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
329 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
330 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
331 // CPP-CHECK-NEXT: ret void
332 //
333 void test_svst2q_vnum_s16(svbool_t pg, const int16_t *base, int64_t vnum, svint16x2_t zt) ATTR
334 {
335 SVE_ACLE_FUNC(svst2q_vnum,,_s16,)(pg, base, vnum, zt);
336 }
337
338 // CHECK-LABEL: @test_svst2q_vnum_u32(
339 // CHECK-NEXT: entry:
340 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
341 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
342 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
343 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
344 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
345 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
346 // CHECK-NEXT: ret void
347 //
348 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_u32u10__SVBool_tPKjl12svuint32x2_t(
349 // CPP-CHECK-NEXT: entry:
350 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
351 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
352 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
353 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
354 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
355 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
356 // CPP-CHECK-NEXT: ret void
357 //
358 void test_svst2q_vnum_u32(svbool_t pg, const uint32_t *base, int64_t vnum, svuint32x2_t zt) ATTR
359 {
360 SVE_ACLE_FUNC(svst2q_vnum,,_u32,)(pg, base, vnum, zt);
361 }
362
363 // CHECK-LABEL: @test_svst2q_vnum_s32(
364 // CHECK-NEXT: entry:
365 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
366 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
367 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
368 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
369 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
370 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
371 // CHECK-NEXT: ret void
372 //
373 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_s32u10__SVBool_tPKil11svint32x2_t(
374 // CPP-CHECK-NEXT: entry:
375 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
376 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
377 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
378 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
379 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
380 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
381 // CPP-CHECK-NEXT: ret void
382 //
383 void test_svst2q_vnum_s32(svbool_t pg, const int32_t *base, int64_t vnum, svint32x2_t zt) ATTR
384 {
385 SVE_ACLE_FUNC(svst2q_vnum,,_s32,)(pg, base, vnum, zt);
386 }
387
388 // CHECK-LABEL: @test_svst2q_vnum_u64(
389 // CHECK-NEXT: entry:
390 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
391 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
392 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
393 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
394 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
395 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
396 // CHECK-NEXT: ret void
397 //
398 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_u64u10__SVBool_tPKml12svuint64x2_t(
399 // CPP-CHECK-NEXT: entry:
400 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
401 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
402 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
403 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
404 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
405 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
406 // CPP-CHECK-NEXT: ret void
407 //
408 void test_svst2q_vnum_u64(svbool_t pg, const uint64_t *base, int64_t vnum, svuint64x2_t zt) ATTR
409 {
410 SVE_ACLE_FUNC(svst2q_vnum,,_u64,)(pg, base, vnum, zt);
411 }
412
413 // CHECK-LABEL: @test_svst2q_vnum_s64(
414 // CHECK-NEXT: entry:
415 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
416 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
417 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
418 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
419 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
420 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
421 // CHECK-NEXT: ret void
422 //
423 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_s64u10__SVBool_tPKll11svint64x2_t(
424 // CPP-CHECK-NEXT: entry:
425 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
426 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
427 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
428 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
429 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
430 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
431 // CPP-CHECK-NEXT: ret void
432 //
433 void test_svst2q_vnum_s64(svbool_t pg, const int64_t *base, int64_t vnum, svint64x2_t zt) ATTR
434 {
435 SVE_ACLE_FUNC(svst2q_vnum,,_s64,)(pg, base, vnum, zt);
436 }
437
438 // CHECK-LABEL: @test_svst2q_vnum_f16(
439 // CHECK-NEXT: entry:
440 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
441 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
442 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
443 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
444 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
445 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
446 // CHECK-NEXT: ret void
447 //
448 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_f16u10__SVBool_tPKDhl13svfloat16x2_t(
449 // CPP-CHECK-NEXT: entry:
450 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
451 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
452 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
453 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
454 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
455 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
456 // CPP-CHECK-NEXT: ret void
457 //
458 void test_svst2q_vnum_f16(svbool_t pg, const float16_t *base, int64_t vnum, svfloat16x2_t zt) ATTR
459 {
460 SVE_ACLE_FUNC(svst2q_vnum,,_f16,)(pg, base, vnum, zt);
461 }
462
463 // CHECK-LABEL: @test_svst2q_vnum_bf16(
464 // CHECK-NEXT: entry:
465 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
466 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
467 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
468 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
469 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
470 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
471 // CHECK-NEXT: ret void
472 //
473 // CPP-CHECK-LABEL: @_Z21test_svst2q_vnum_bf16u10__SVBool_tPKu6__bf16l14svbfloat16x2_t(
474 // CPP-CHECK-NEXT: entry:
475 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
476 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
477 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
478 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
479 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
480 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
481 // CPP-CHECK-NEXT: ret void
482 //
483 void test_svst2q_vnum_bf16(svbool_t pg, const bfloat16_t *base, int64_t vnum, svbfloat16x2_t zt) ATTR
484 {
485 SVE_ACLE_FUNC(svst2q_vnum,,_bf16,)(pg, base, vnum, zt);
486 }
487
488 // CHECK-LABEL: @test_svst2q_vnum_f32(
489 // CHECK-NEXT: entry:
490 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
491 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
492 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
493 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
494 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
495 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
496 // CHECK-NEXT: ret void
497 //
498 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_f32u10__SVBool_tPKfl13svfloat32x2_t(
499 // CPP-CHECK-NEXT: entry:
500 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
501 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
502 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
503 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
504 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
505 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
506 // CPP-CHECK-NEXT: ret void
507 //
508 void test_svst2q_vnum_f32(svbool_t pg, const float32_t *base, int64_t vnum, svfloat32x2_t zt) ATTR
509 {
510 SVE_ACLE_FUNC(svst2q_vnum,,_f32,)(pg, base, vnum, zt);
511 }
512
513 // CHECK-LABEL: @test_svst2q_vnum_f64(
514 // CHECK-NEXT: entry:
515 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
516 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
517 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
518 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
519 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
520 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
521 // CHECK-NEXT: ret void
522 //
523 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_f64u10__SVBool_tPKdl13svfloat64x2_t(
524 // CPP-CHECK-NEXT: entry:
525 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
526 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
527 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
528 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
529 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
530 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
531 // CPP-CHECK-NEXT: ret void
532 //
533 void test_svst2q_vnum_f64(svbool_t pg, const float64_t *base, int64_t vnum, svfloat64x2_t zt) ATTR
534 {
535 SVE_ACLE_FUNC(svst2q_vnum,,_f64,)(pg, base, vnum, zt);
536 }
537
538 // CHECK-LABEL: @test_svst2q_vnum_mf8(
539 // CHECK-NEXT: entry:
540 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
541 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
542 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
543 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
544 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
545 // CHECK-NEXT: ret void
546 //
547 // CPP-CHECK-LABEL: @_Z20test_svst2q_vnum_mf8u10__SVBool_tPKu6__mfp8l13svmfloat8x2_t(
548 // CPP-CHECK-NEXT: entry:
549 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
550 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
551 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
552 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
553 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
554 // CPP-CHECK-NEXT: ret void
555 //
556 void test_svst2q_vnum_mf8(svbool_t pg, const mfloat8_t *base, int64_t vnum, svmfloat8x2_t zt) ATTR
557 {
558 SVE_ACLE_FUNC(svst2q_vnum,,_mf8,)(pg, base, vnum, zt);
559 }
560
561 //
562 // ST3Q
563 // CHECK-LABEL: @test_svst3q_u8(
564 // CHECK-NEXT: entry:
565 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
566 // CHECK-NEXT: ret void
567 //
568 // CPP-CHECK-LABEL: @_Z14test_svst3q_u8u10__SVBool_tPKh11svuint8x3_t(
569 // CPP-CHECK-NEXT: entry:
570 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
571 // CPP-CHECK-NEXT: ret void
572 //
573 void test_svst3q_u8(svbool_t pg, const uint8_t *base, svuint8x3_t zt) ATTR
574 {
575 SVE_ACLE_FUNC(svst3q,,_u8,)(pg, base, zt);
576 }
577
578 // CHECK-LABEL: @test_svst3q_s8(
579 // CHECK-NEXT: entry:
580 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
581 // CHECK-NEXT: ret void
582 //
583 // CPP-CHECK-LABEL: @_Z14test_svst3q_s8u10__SVBool_tPKa10svint8x3_t(
584 // CPP-CHECK-NEXT: entry:
585 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
586 // CPP-CHECK-NEXT: ret void
587 //
588 void test_svst3q_s8(svbool_t pg, const int8_t *base, svint8x3_t zt) ATTR
589 {
590 SVE_ACLE_FUNC(svst3q,,_s8,)(pg, base, zt);
591 }
592 // CHECK-LABEL: @test_svst3q_u16(
593 // CHECK-NEXT: entry:
594 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
595 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
596 // CHECK-NEXT: ret void
597 //
598 // CPP-CHECK-LABEL: @_Z15test_svst3q_u16u10__SVBool_tPKt12svuint16x3_t(
599 // CPP-CHECK-NEXT: entry:
600 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
601 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
602 // CPP-CHECK-NEXT: ret void
603 //
604 void test_svst3q_u16(svbool_t pg, const uint16_t *base, svuint16x3_t zt) ATTR
605 {
606 SVE_ACLE_FUNC(svst3q,,_u16,)(pg, base, zt);
607 }
608
609 // CHECK-LABEL: @test_svst3q_s16(
610 // CHECK-NEXT: entry:
611 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
612 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
613 // CHECK-NEXT: ret void
614 //
615 // CPP-CHECK-LABEL: @_Z15test_svst3q_s16u10__SVBool_tPKs11svint16x3_t(
616 // CPP-CHECK-NEXT: entry:
617 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
618 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
619 // CPP-CHECK-NEXT: ret void
620 //
621 void test_svst3q_s16(svbool_t pg, const int16_t *base, svint16x3_t zt) ATTR
622 {
623 SVE_ACLE_FUNC(svst3q,,_s16,)(pg, base, zt);
624 }
625
626 // CHECK-LABEL: @test_svst3q_u32(
627 // CHECK-NEXT: entry:
628 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
629 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
630 // CHECK-NEXT: ret void
631 //
632 // CPP-CHECK-LABEL: @_Z15test_svst3q_u32u10__SVBool_tPKj12svuint32x3_t(
633 // CPP-CHECK-NEXT: entry:
634 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
635 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
636 // CPP-CHECK-NEXT: ret void
637 //
638 void test_svst3q_u32(svbool_t pg, const uint32_t *base, svuint32x3_t zt) ATTR
639 {
640 SVE_ACLE_FUNC(svst3q,,_u32,)(pg, base, zt);
641 }
642
643 // CHECK-LABEL: @test_svst3q_s32(
644 // CHECK-NEXT: entry:
645 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
646 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
647 // CHECK-NEXT: ret void
648 //
649 // CPP-CHECK-LABEL: @_Z15test_svst3q_s32u10__SVBool_tPKi11svint32x3_t(
650 // CPP-CHECK-NEXT: entry:
651 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
652 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
653 // CPP-CHECK-NEXT: ret void
654 //
655 void test_svst3q_s32(svbool_t pg, const int32_t *base, svint32x3_t zt) ATTR
656 {
657 SVE_ACLE_FUNC(svst3q,,_s32,)(pg, base, zt);
658 }
659
660 // CHECK-LABEL: @test_svst3q_u64(
661 // CHECK-NEXT: entry:
662 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
663 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
664 // CHECK-NEXT: ret void
665 //
666 // CPP-CHECK-LABEL: @_Z15test_svst3q_u64u10__SVBool_tPKm12svuint64x3_t(
667 // CPP-CHECK-NEXT: entry:
668 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
669 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
670 // CPP-CHECK-NEXT: ret void
671 //
672 void test_svst3q_u64(svbool_t pg, const uint64_t *base, svuint64x3_t zt) ATTR
673 {
674 SVE_ACLE_FUNC(svst3q,,_u64,)(pg, base, zt);
675 }
676
677 // CHECK-LABEL: @test_svst3q_s64(
678 // CHECK-NEXT: entry:
679 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
680 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
681 // CHECK-NEXT: ret void
682 //
683 // CPP-CHECK-LABEL: @_Z15test_svst3q_s64u10__SVBool_tPKl11svint64x3_t(
684 // CPP-CHECK-NEXT: entry:
685 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
686 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
687 // CPP-CHECK-NEXT: ret void
688 //
689 void test_svst3q_s64(svbool_t pg, const int64_t *base, svint64x3_t zt) ATTR
690 {
691 SVE_ACLE_FUNC(svst3q,,_s64,)(pg, base, zt);
692 }
693
694 // CHECK-LABEL: @test_svst3q_f16(
695 // CHECK-NEXT: entry:
696 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
697 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
698 // CHECK-NEXT: ret void
699 //
700 // CPP-CHECK-LABEL: @_Z15test_svst3q_f16u10__SVBool_tPKDh13svfloat16x3_t(
701 // CPP-CHECK-NEXT: entry:
702 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
703 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
704 // CPP-CHECK-NEXT: ret void
705 //
706 void test_svst3q_f16(svbool_t pg, const float16_t *base, svfloat16x3_t zt) ATTR
707 {
708 SVE_ACLE_FUNC(svst3q,,_f16,)(pg, base, zt);
709 }
710
711 // CHECK-LABEL: @test_svst3q_bf16(
712 // CHECK-NEXT: entry:
713 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
714 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
715 // CHECK-NEXT: ret void
716 //
717 // CPP-CHECK-LABEL: @_Z16test_svst3q_bf16u10__SVBool_tPKu6__bf1614svbfloat16x3_t(
718 // CPP-CHECK-NEXT: entry:
719 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
720 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
721 // CPP-CHECK-NEXT: ret void
722 //
723 void test_svst3q_bf16(svbool_t pg, const bfloat16_t *base, svbfloat16x3_t zt) ATTR
724 {
725 SVE_ACLE_FUNC(svst3q,,_bf16,)(pg, base, zt);
726 }
727
728 // CHECK-LABEL: @test_svst3q_f32(
729 // CHECK-NEXT: entry:
730 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
731 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
732 // CHECK-NEXT: ret void
733 //
734 // CPP-CHECK-LABEL: @_Z15test_svst3q_f32u10__SVBool_tPKf13svfloat32x3_t(
735 // CPP-CHECK-NEXT: entry:
736 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
737 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
738 // CPP-CHECK-NEXT: ret void
739 //
740 void test_svst3q_f32(svbool_t pg, const float32_t *base, svfloat32x3_t zt) ATTR
741 {
742 SVE_ACLE_FUNC(svst3q,,_f32,)(pg, base, zt);
743 }
744
745 // CHECK-LABEL: @test_svst3q_f64(
746 // CHECK-NEXT: entry:
747 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
748 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
749 // CHECK-NEXT: ret void
750 //
751 // CPP-CHECK-LABEL: @_Z15test_svst3q_f64u10__SVBool_tPKd13svfloat64x3_t(
752 // CPP-CHECK-NEXT: entry:
753 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
754 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
755 // CPP-CHECK-NEXT: ret void
756 //
757 void test_svst3q_f64(svbool_t pg, const float64_t *base, svfloat64x3_t zt) ATTR
758 {
759 SVE_ACLE_FUNC(svst3q,,_f64,)(pg, base, zt);
760 }
761
762 // CHECK-LABEL: @test_svst3q_mf8(
763 // CHECK-NEXT: entry:
764 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
765 // CHECK-NEXT: ret void
766 //
767 // CPP-CHECK-LABEL: @_Z15test_svst3q_mf8u10__SVBool_tPKu6__mfp813svmfloat8x3_t(
768 // CPP-CHECK-NEXT: entry:
769 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
770 // CPP-CHECK-NEXT: ret void
771 //
772 void test_svst3q_mf8(svbool_t pg, const mfloat8_t *base, svmfloat8x3_t zt) ATTR
773 {
774 SVE_ACLE_FUNC(svst3q,,_mf8,)(pg, base, zt);
775 }
776
777 // CHECK-LABEL: @test_svst3q_vnum_u8(
778 // CHECK-NEXT: entry:
779 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
780 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
781 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
782 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
783 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
784 // CHECK-NEXT: ret void
785 //
786 // CPP-CHECK-LABEL: @_Z19test_svst3q_vnum_u8u10__SVBool_tPKhl11svuint8x3_t(
787 // CPP-CHECK-NEXT: entry:
788 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
789 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
790 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
791 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
792 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
793 // CPP-CHECK-NEXT: ret void
794 //
795 void test_svst3q_vnum_u8(svbool_t pg, const uint8_t *base, int64_t vnum, svuint8x3_t zt) ATTR
796 {
797 SVE_ACLE_FUNC(svst3q_vnum_,,u8,)(pg, base, vnum, zt);
798 }
799
800 // CHECK-LABEL: @test_svst3q_vnum_s8(
801 // CHECK-NEXT: entry:
802 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
803 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
804 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
805 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
806 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
807 // CHECK-NEXT: ret void
808 //
809 // CPP-CHECK-LABEL: @_Z19test_svst3q_vnum_s8u10__SVBool_tPKal10svint8x3_t(
810 // CPP-CHECK-NEXT: entry:
811 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
812 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
813 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
814 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
815 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
816 // CPP-CHECK-NEXT: ret void
817 //
818 void test_svst3q_vnum_s8(svbool_t pg, const int8_t *base, int64_t vnum, svint8x3_t zt) ATTR
819 {
820 SVE_ACLE_FUNC(svst3q_vnum,,_s8,)(pg, base, vnum, zt);
821 }
822
823 // CHECK-LABEL: @test_svst3q_vnum_u16(
824 // CHECK-NEXT: entry:
825 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
826 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
827 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
828 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
829 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
830 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
831 // CHECK-NEXT: ret void
832 //
833 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_u16u10__SVBool_tPKtl12svuint16x3_t(
834 // CPP-CHECK-NEXT: entry:
835 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
836 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
837 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
838 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
839 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
840 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
841 // CPP-CHECK-NEXT: ret void
842 //
843 void test_svst3q_vnum_u16(svbool_t pg, const uint16_t *base, int64_t vnum, svuint16x3_t zt) ATTR
844 {
845 SVE_ACLE_FUNC(svst3q_vnum,,_u16,)(pg, base, vnum, zt);
846 }
847
848 // CHECK-LABEL: @test_svst3q_vnum_s16(
849 // CHECK-NEXT: entry:
850 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
851 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
852 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
853 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
854 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
855 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
856 // CHECK-NEXT: ret void
857 //
858 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_s16u10__SVBool_tPKsl11svint16x3_t(
859 // CPP-CHECK-NEXT: entry:
860 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
861 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
862 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
863 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
864 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
865 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
866 // CPP-CHECK-NEXT: ret void
867 //
868 void test_svst3q_vnum_s16(svbool_t pg, const int16_t *base, int64_t vnum, svint16x3_t zt) ATTR
869 {
870 SVE_ACLE_FUNC(svst3q_vnum,,_s16,)(pg, base, vnum, zt);
871 }
872
873 // CHECK-LABEL: @test_svst3q_vnum_u32(
874 // CHECK-NEXT: entry:
875 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
876 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
877 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
878 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
879 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
880 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
881 // CHECK-NEXT: ret void
882 //
883 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_u32u10__SVBool_tPKjl12svuint32x3_t(
884 // CPP-CHECK-NEXT: entry:
885 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
886 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
887 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
888 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
889 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
890 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
891 // CPP-CHECK-NEXT: ret void
892 //
893 void test_svst3q_vnum_u32(svbool_t pg, const uint32_t *base, int64_t vnum, svuint32x3_t zt) ATTR
894 {
895 SVE_ACLE_FUNC(svst3q_vnum,,_u32,)(pg, base, vnum, zt);
896 }
897
898 // CHECK-LABEL: @test_svst3q_vnum_s32(
899 // CHECK-NEXT: entry:
900 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
901 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
902 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
903 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
904 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
905 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
906 // CHECK-NEXT: ret void
907 //
908 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_s32u10__SVBool_tPKil11svint32x3_t(
909 // CPP-CHECK-NEXT: entry:
910 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
911 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
912 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
913 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
914 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
915 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
916 // CPP-CHECK-NEXT: ret void
917 //
918 void test_svst3q_vnum_s32(svbool_t pg, const int32_t *base, int64_t vnum, svint32x3_t zt) ATTR
919 {
920 SVE_ACLE_FUNC(svst3q_vnum,,_s32,)(pg, base, vnum, zt);
921 }
922
923 // CHECK-LABEL: @test_svst3q_vnum_u64(
924 // CHECK-NEXT: entry:
925 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
926 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
927 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
928 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
929 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
930 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
931 // CHECK-NEXT: ret void
932 //
933 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_u64u10__SVBool_tPKml12svuint64x3_t(
934 // CPP-CHECK-NEXT: entry:
935 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
936 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
937 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
938 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
939 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
940 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
941 // CPP-CHECK-NEXT: ret void
942 //
943 void test_svst3q_vnum_u64(svbool_t pg, const uint64_t *base, int64_t vnum, svuint64x3_t zt) ATTR
944 {
945 SVE_ACLE_FUNC(svst3q_vnum,,_u64,)(pg, base, vnum, zt);
946 }
947
948 // CHECK-LABEL: @test_svst3q_vnum_s64(
949 // CHECK-NEXT: entry:
950 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
951 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
952 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
953 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
954 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
955 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
956 // CHECK-NEXT: ret void
957 //
958 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_s64u10__SVBool_tPKll11svint64x3_t(
959 // CPP-CHECK-NEXT: entry:
960 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
961 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
962 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
963 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
964 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
965 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
966 // CPP-CHECK-NEXT: ret void
967 //
968 void test_svst3q_vnum_s64(svbool_t pg, const int64_t *base, int64_t vnum, svint64x3_t zt) ATTR
969 {
970 SVE_ACLE_FUNC(svst3q_vnum,,_s64,)(pg, base, vnum, zt);
971 }
972
973 // CHECK-LABEL: @test_svst3q_vnum_f16(
974 // CHECK-NEXT: entry:
975 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
976 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
977 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
978 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
979 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
980 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
981 // CHECK-NEXT: ret void
982 //
983 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_f16u10__SVBool_tPKDhl13svfloat16x3_t(
984 // CPP-CHECK-NEXT: entry:
985 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
986 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
987 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
988 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
989 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
990 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
991 // CPP-CHECK-NEXT: ret void
992 //
993 void test_svst3q_vnum_f16(svbool_t pg, const float16_t *base, int64_t vnum, svfloat16x3_t zt) ATTR
994 {
995 SVE_ACLE_FUNC(svst3q_vnum,,_f16,)(pg, base, vnum, zt);
996 }
997
998 // CHECK-LABEL: @test_svst3q_vnum_bf16(
999 // CHECK-NEXT: entry:
1000 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1001 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1002 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1003 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1004 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1005 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1006 // CHECK-NEXT: ret void
1007 //
1008 // CPP-CHECK-LABEL: @_Z21test_svst3q_vnum_bf16u10__SVBool_tPKu6__bf16l14svbfloat16x3_t(
1009 // CPP-CHECK-NEXT: entry:
1010 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1011 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1012 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1013 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1014 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1015 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1016 // CPP-CHECK-NEXT: ret void
1017 //
1018 void test_svst3q_vnum_bf16(svbool_t pg, const bfloat16_t *base, int64_t vnum, svbfloat16x3_t zt) ATTR
1019 {
1020 SVE_ACLE_FUNC(svst3q_vnum,,_bf16,)(pg, base, vnum, zt);
1021 }
1022
1023 // CHECK-LABEL: @test_svst3q_vnum_f32(
1024 // CHECK-NEXT: entry:
1025 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1026 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1027 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1028 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1029 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1030 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1031 // CHECK-NEXT: ret void
1032 //
1033 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_f32u10__SVBool_tPKfl13svfloat32x3_t(
1034 // CPP-CHECK-NEXT: entry:
1035 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1036 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1037 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1038 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1039 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1040 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1041 // CPP-CHECK-NEXT: ret void
1042 //
1043 void test_svst3q_vnum_f32(svbool_t pg, const float32_t *base, int64_t vnum, svfloat32x3_t zt) ATTR
1044 {
1045 SVE_ACLE_FUNC(svst3q_vnum,,_f32,)(pg, base, vnum, zt);
1046 }
1047
1048 // CHECK-LABEL: @test_svst3q_vnum_f64(
1049 // CHECK-NEXT: entry:
1050 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1051 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1052 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1053 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1054 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1055 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1056 // CHECK-NEXT: ret void
1057 //
1058 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_f64u10__SVBool_tPKdl13svfloat64x3_t(
1059 // CPP-CHECK-NEXT: entry:
1060 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1061 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1062 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1063 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1064 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1065 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1066 // CPP-CHECK-NEXT: ret void
1067 //
1068 void test_svst3q_vnum_f64(svbool_t pg, const float64_t *base, int64_t vnum, svfloat64x3_t zt) ATTR
1069 {
1070 SVE_ACLE_FUNC(svst3q_vnum,,_f64,)(pg, base, vnum, zt);
1071 }
1072
1073 // CHECK-LABEL: @test_svst3q_vnum_mf8(
1074 // CHECK-NEXT: entry:
1075 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1076 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1077 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1078 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1079 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1080 // CHECK-NEXT: ret void
1081 //
1082 // CPP-CHECK-LABEL: @_Z20test_svst3q_vnum_mf8u10__SVBool_tPKu6__mfp8l13svmfloat8x3_t(
1083 // CPP-CHECK-NEXT: entry:
1084 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1085 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1086 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1087 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1088 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st3q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1089 // CPP-CHECK-NEXT: ret void
1090 //
1091 void test_svst3q_vnum_mf8(svbool_t pg, const mfloat8_t *base, int64_t vnum, svmfloat8x3_t zt) ATTR
1092 {
1093 SVE_ACLE_FUNC(svst3q_vnum,,_mf8,)(pg, base, vnum, zt);
1094 }
1095
1096 //
1097 // ST4Q
1098 // CHECK-LABEL: @test_svst4q_u8(
1099 // CHECK-NEXT: entry:
1100 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
1101 // CHECK-NEXT: ret void
1102 //
1103 // CPP-CHECK-LABEL: @_Z14test_svst4q_u8u10__SVBool_tPKh11svuint8x4_t(
1104 // CPP-CHECK-NEXT: entry:
1105 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
1106 // CPP-CHECK-NEXT: ret void
1107 //
1108 void test_svst4q_u8(svbool_t pg, const uint8_t *base, svuint8x4_t zt) ATTR
1109 {
1110 SVE_ACLE_FUNC(svst4q,,_u8,)(pg, base, zt);
1111 }
1112
1113 // CHECK-LABEL: @test_svst4q_s8(
1114 // CHECK-NEXT: entry:
1115 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
1116 // CHECK-NEXT: ret void
1117 //
1118 // CPP-CHECK-LABEL: @_Z14test_svst4q_s8u10__SVBool_tPKa10svint8x4_t(
1119 // CPP-CHECK-NEXT: entry:
1120 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
1121 // CPP-CHECK-NEXT: ret void
1122 //
1123 void test_svst4q_s8(svbool_t pg, const int8_t *base, svint8x4_t zt) ATTR
1124 {
1125 SVE_ACLE_FUNC(svst4q,,_s8,)(pg, base, zt);
1126 }
1127 // CHECK-LABEL: @test_svst4q_u16(
1128 // CHECK-NEXT: entry:
1129 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1130 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1131 // CHECK-NEXT: ret void
1132 //
1133 // CPP-CHECK-LABEL: @_Z15test_svst4q_u16u10__SVBool_tPKt12svuint16x4_t(
1134 // CPP-CHECK-NEXT: entry:
1135 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1136 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1137 // CPP-CHECK-NEXT: ret void
1138 //
1139 void test_svst4q_u16(svbool_t pg, const uint16_t *base, svuint16x4_t zt) ATTR
1140 {
1141 SVE_ACLE_FUNC(svst4q,,_u16,)(pg, base, zt);
1142 }
1143
1144 // CHECK-LABEL: @test_svst4q_s16(
1145 // CHECK-NEXT: entry:
1146 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1147 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1148 // CHECK-NEXT: ret void
1149 //
1150 // CPP-CHECK-LABEL: @_Z15test_svst4q_s16u10__SVBool_tPKs11svint16x4_t(
1151 // CPP-CHECK-NEXT: entry:
1152 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1153 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1154 // CPP-CHECK-NEXT: ret void
1155 //
1156 void test_svst4q_s16(svbool_t pg, const int16_t *base, svint16x4_t zt) ATTR
1157 {
1158 SVE_ACLE_FUNC(svst4q,,_s16,)(pg, base, zt);
1159 }
1160
1161 // CHECK-LABEL: @test_svst4q_u32(
1162 // CHECK-NEXT: entry:
1163 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1164 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
1165 // CHECK-NEXT: ret void
1166 //
1167 // CPP-CHECK-LABEL: @_Z15test_svst4q_u32u10__SVBool_tPKj12svuint32x4_t(
1168 // CPP-CHECK-NEXT: entry:
1169 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1170 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
1171 // CPP-CHECK-NEXT: ret void
1172 //
1173 void test_svst4q_u32(svbool_t pg, const uint32_t *base, svuint32x4_t zt) ATTR
1174 {
1175 SVE_ACLE_FUNC(svst4q,,_u32,)(pg, base, zt);
1176 }
1177
1178 // CHECK-LABEL: @test_svst4q_s32(
1179 // CHECK-NEXT: entry:
1180 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1181 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
1182 // CHECK-NEXT: ret void
1183 //
1184 // CPP-CHECK-LABEL: @_Z15test_svst4q_s32u10__SVBool_tPKi11svint32x4_t(
1185 // CPP-CHECK-NEXT: entry:
1186 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1187 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
1188 // CPP-CHECK-NEXT: ret void
1189 //
1190 void test_svst4q_s32(svbool_t pg, const int32_t *base, svint32x4_t zt) ATTR
1191 {
1192 SVE_ACLE_FUNC(svst4q,,_s32,)(pg, base, zt);
1193 }
1194
1195 // CHECK-LABEL: @test_svst4q_u64(
1196 // CHECK-NEXT: entry:
1197 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1198 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
1199 // CHECK-NEXT: ret void
1200 //
1201 // CPP-CHECK-LABEL: @_Z15test_svst4q_u64u10__SVBool_tPKm12svuint64x4_t(
1202 // CPP-CHECK-NEXT: entry:
1203 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1204 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
1205 // CPP-CHECK-NEXT: ret void
1206 //
1207 void test_svst4q_u64(svbool_t pg, const uint64_t *base, svuint64x4_t zt) ATTR
1208 {
1209 SVE_ACLE_FUNC(svst4q,,_u64,)(pg, base, zt);
1210 }
1211
1212 // CHECK-LABEL: @test_svst4q_s64(
1213 // CHECK-NEXT: entry:
1214 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1215 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
1216 // CHECK-NEXT: ret void
1217 //
1218 // CPP-CHECK-LABEL: @_Z15test_svst4q_s64u10__SVBool_tPKl11svint64x4_t(
1219 // CPP-CHECK-NEXT: entry:
1220 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1221 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
1222 // CPP-CHECK-NEXT: ret void
1223 //
1224 void test_svst4q_s64(svbool_t pg, const int64_t *base, svint64x4_t zt) ATTR
1225 {
1226 SVE_ACLE_FUNC(svst4q,,_s64,)(pg, base, zt);
1227 }
1228
1229 // CHECK-LABEL: @test_svst4q_f16(
1230 // CHECK-NEXT: entry:
1231 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1232 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x half> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1233 // CHECK-NEXT: ret void
1234 //
1235 // CPP-CHECK-LABEL: @_Z15test_svst4q_f16u10__SVBool_tPKDh13svfloat16x4_t(
1236 // CPP-CHECK-NEXT: entry:
1237 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1238 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x half> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1239 // CPP-CHECK-NEXT: ret void
1240 //
1241 void test_svst4q_f16(svbool_t pg, const float16_t *base, svfloat16x4_t zt) ATTR
1242 {
1243 SVE_ACLE_FUNC(svst4q,,_f16,)(pg, base, zt);
1244 }
1245
1246 // CHECK-LABEL: @test_svst4q_bf16(
1247 // CHECK-NEXT: entry:
1248 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1249 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1250 // CHECK-NEXT: ret void
1251 //
1252 // CPP-CHECK-LABEL: @_Z16test_svst4q_bf16u10__SVBool_tPKu6__bf1614svbfloat16x4_t(
1253 // CPP-CHECK-NEXT: entry:
1254 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1255 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
1256 // CPP-CHECK-NEXT: ret void
1257 //
1258 void test_svst4q_bf16(svbool_t pg, const bfloat16_t *base, svbfloat16x4_t zt) ATTR
1259 {
1260 SVE_ACLE_FUNC(svst4q,,_bf16,)(pg, base, zt);
1261 }
1262
1263 // CHECK-LABEL: @test_svst4q_f32(
1264 // CHECK-NEXT: entry:
1265 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1266 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x float> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
1267 // CHECK-NEXT: ret void
1268 //
1269 // CPP-CHECK-LABEL: @_Z15test_svst4q_f32u10__SVBool_tPKf13svfloat32x4_t(
1270 // CPP-CHECK-NEXT: entry:
1271 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1272 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x float> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
1273 // CPP-CHECK-NEXT: ret void
1274 //
1275 void test_svst4q_f32(svbool_t pg, const float32_t *base, svfloat32x4_t zt) ATTR
1276 {
1277 SVE_ACLE_FUNC(svst4q,,_f32,)(pg, base, zt);
1278 }
1279
1280 // CHECK-LABEL: @test_svst4q_f64(
1281 // CHECK-NEXT: entry:
1282 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1283 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x double> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
1284 // CHECK-NEXT: ret void
1285 //
1286 // CPP-CHECK-LABEL: @_Z15test_svst4q_f64u10__SVBool_tPKd13svfloat64x4_t(
1287 // CPP-CHECK-NEXT: entry:
1288 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1289 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x double> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
1290 // CPP-CHECK-NEXT: ret void
1291 //
1292 void test_svst4q_f64(svbool_t pg, const float64_t *base, svfloat64x4_t zt) ATTR
1293 {
1294 SVE_ACLE_FUNC(svst4q,,_f64,)(pg, base, zt);
1295 }
1296
1297 // CHECK-LABEL: @test_svst4q_mf8(
1298 // CHECK-NEXT: entry:
1299 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
1300 // CHECK-NEXT: ret void
1301 //
1302 // CPP-CHECK-LABEL: @_Z15test_svst4q_mf8u10__SVBool_tPKu6__mfp813svmfloat8x4_t(
1303 // CPP-CHECK-NEXT: entry:
1304 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
1305 // CPP-CHECK-NEXT: ret void
1306 //
1307 void test_svst4q_mf8(svbool_t pg, const mfloat8_t *base, svmfloat8x4_t zt) ATTR
1308 {
1309 SVE_ACLE_FUNC(svst4q,,_mf8,)(pg, base, zt);
1310 }
1311
1312 // CHECK-LABEL: @test_svst4q_vnum_u8(
1313 // CHECK-NEXT: entry:
1314 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1315 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1316 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1317 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1318 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1319 // CHECK-NEXT: ret void
1320 //
1321 // CPP-CHECK-LABEL: @_Z19test_svst4q_vnum_u8u10__SVBool_tPKhl11svuint8x4_t(
1322 // CPP-CHECK-NEXT: entry:
1323 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1324 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1325 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1326 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1327 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1328 // CPP-CHECK-NEXT: ret void
1329 //
1330 void test_svst4q_vnum_u8(svbool_t pg, const uint8_t *base, int64_t vnum, svuint8x4_t zt) ATTR
1331 {
1332 SVE_ACLE_FUNC(svst4q_vnum_,,u8,)(pg, base, vnum, zt);
1333 }
1334
1335 // CHECK-LABEL: @test_svst4q_vnum_s8(
1336 // CHECK-NEXT: entry:
1337 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1338 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1339 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1340 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1341 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1342 // CHECK-NEXT: ret void
1343 //
1344 // CPP-CHECK-LABEL: @_Z19test_svst4q_vnum_s8u10__SVBool_tPKal10svint8x4_t(
1345 // CPP-CHECK-NEXT: entry:
1346 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1347 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1348 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1349 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1350 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1351 // CPP-CHECK-NEXT: ret void
1352 //
1353 void test_svst4q_vnum_s8(svbool_t pg, const int8_t *base, int64_t vnum, svint8x4_t zt) ATTR
1354 {
1355 SVE_ACLE_FUNC(svst4q_vnum,,_s8,)(pg, base, vnum, zt);
1356 }
1357
1358 // CHECK-LABEL: @test_svst4q_vnum_u16(
1359 // CHECK-NEXT: entry:
1360 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1361 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1362 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1363 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1364 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1365 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1366 // CHECK-NEXT: ret void
1367 //
1368 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_u16u10__SVBool_tPKtl12svuint16x4_t(
1369 // CPP-CHECK-NEXT: entry:
1370 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1371 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1372 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1373 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1374 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1375 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1376 // CPP-CHECK-NEXT: ret void
1377 //
1378 void test_svst4q_vnum_u16(svbool_t pg, const uint16_t *base, int64_t vnum, svuint16x4_t zt) ATTR
1379 {
1380 SVE_ACLE_FUNC(svst4q_vnum,,_u16,)(pg, base, vnum, zt);
1381 }
1382
1383 // CHECK-LABEL: @test_svst4q_vnum_s16(
1384 // CHECK-NEXT: entry:
1385 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1386 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1387 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1388 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1389 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1390 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1391 // CHECK-NEXT: ret void
1392 //
1393 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_s16u10__SVBool_tPKsl11svint16x4_t(
1394 // CPP-CHECK-NEXT: entry:
1395 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1396 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1397 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1398 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1399 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1400 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8i16(<vscale x 8 x i16> [[ZT_COERCE0:%.*]], <vscale x 8 x i16> [[ZT_COERCE1:%.*]], <vscale x 8 x i16> [[ZT_COERCE2:%.*]], <vscale x 8 x i16> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1401 // CPP-CHECK-NEXT: ret void
1402 //
1403 void test_svst4q_vnum_s16(svbool_t pg, const int16_t *base, int64_t vnum, svint16x4_t zt) ATTR
1404 {
1405 SVE_ACLE_FUNC(svst4q_vnum,,_s16,)(pg, base, vnum, zt);
1406 }
1407
1408 // CHECK-LABEL: @test_svst4q_vnum_u32(
1409 // CHECK-NEXT: entry:
1410 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1411 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1412 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1413 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1414 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1415 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1416 // CHECK-NEXT: ret void
1417 //
1418 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_u32u10__SVBool_tPKjl12svuint32x4_t(
1419 // CPP-CHECK-NEXT: entry:
1420 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1421 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1422 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1423 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1424 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1425 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1426 // CPP-CHECK-NEXT: ret void
1427 //
1428 void test_svst4q_vnum_u32(svbool_t pg, const uint32_t *base, int64_t vnum, svuint32x4_t zt) ATTR
1429 {
1430 SVE_ACLE_FUNC(svst4q_vnum,,_u32,)(pg, base, vnum, zt);
1431 }
1432
1433 // CHECK-LABEL: @test_svst4q_vnum_s32(
1434 // CHECK-NEXT: entry:
1435 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1436 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1437 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1438 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1439 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1440 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1441 // CHECK-NEXT: ret void
1442 //
1443 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_s32u10__SVBool_tPKil11svint32x4_t(
1444 // CPP-CHECK-NEXT: entry:
1445 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1446 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1447 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1448 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1449 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1450 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4i32(<vscale x 4 x i32> [[ZT_COERCE0:%.*]], <vscale x 4 x i32> [[ZT_COERCE1:%.*]], <vscale x 4 x i32> [[ZT_COERCE2:%.*]], <vscale x 4 x i32> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1451 // CPP-CHECK-NEXT: ret void
1452 //
1453 void test_svst4q_vnum_s32(svbool_t pg, const int32_t *base, int64_t vnum, svint32x4_t zt) ATTR
1454 {
1455 SVE_ACLE_FUNC(svst4q_vnum,,_s32,)(pg, base, vnum, zt);
1456 }
1457
1458 // CHECK-LABEL: @test_svst4q_vnum_u64(
1459 // CHECK-NEXT: entry:
1460 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1461 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1462 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1463 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1464 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1465 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1466 // CHECK-NEXT: ret void
1467 //
1468 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_u64u10__SVBool_tPKml12svuint64x4_t(
1469 // CPP-CHECK-NEXT: entry:
1470 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1471 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1472 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1473 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1474 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1475 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1476 // CPP-CHECK-NEXT: ret void
1477 //
1478 void test_svst4q_vnum_u64(svbool_t pg, const uint64_t *base, int64_t vnum, svuint64x4_t zt) ATTR
1479 {
1480 SVE_ACLE_FUNC(svst4q_vnum,,_u64,)(pg, base, vnum, zt);
1481 }
1482
1483 // CHECK-LABEL: @test_svst4q_vnum_s64(
1484 // CHECK-NEXT: entry:
1485 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1486 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1487 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1488 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1489 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1490 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1491 // CHECK-NEXT: ret void
1492 //
1493 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_s64u10__SVBool_tPKll11svint64x4_t(
1494 // CPP-CHECK-NEXT: entry:
1495 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1496 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1497 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1498 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1499 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1500 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2i64(<vscale x 2 x i64> [[ZT_COERCE0:%.*]], <vscale x 2 x i64> [[ZT_COERCE1:%.*]], <vscale x 2 x i64> [[ZT_COERCE2:%.*]], <vscale x 2 x i64> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1501 // CPP-CHECK-NEXT: ret void
1502 //
1503 void test_svst4q_vnum_s64(svbool_t pg, const int64_t *base, int64_t vnum, svint64x4_t zt) ATTR
1504 {
1505 SVE_ACLE_FUNC(svst4q_vnum,,_s64,)(pg, base, vnum, zt);
1506 }
1507
1508 // CHECK-LABEL: @test_svst4q_vnum_f16(
1509 // CHECK-NEXT: entry:
1510 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1511 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1512 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1513 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1514 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1515 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x half> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1516 // CHECK-NEXT: ret void
1517 //
1518 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_f16u10__SVBool_tPKDhl13svfloat16x4_t(
1519 // CPP-CHECK-NEXT: entry:
1520 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1521 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1522 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1523 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1524 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1525 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8f16(<vscale x 8 x half> [[ZT_COERCE0:%.*]], <vscale x 8 x half> [[ZT_COERCE1:%.*]], <vscale x 8 x half> [[ZT_COERCE2:%.*]], <vscale x 8 x half> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1526 // CPP-CHECK-NEXT: ret void
1527 //
1528 void test_svst4q_vnum_f16(svbool_t pg, const float16_t *base, int64_t vnum, svfloat16x4_t zt) ATTR
1529 {
1530 SVE_ACLE_FUNC(svst4q_vnum,,_f16,)(pg, base, vnum, zt);
1531 }
1532
1533 // CHECK-LABEL: @test_svst4q_vnum_bf16(
1534 // CHECK-NEXT: entry:
1535 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1536 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1537 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1538 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1539 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1540 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1541 // CHECK-NEXT: ret void
1542 //
1543 // CPP-CHECK-LABEL: @_Z21test_svst4q_vnum_bf16u10__SVBool_tPKu6__bf16l14svbfloat16x4_t(
1544 // CPP-CHECK-NEXT: entry:
1545 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
1546 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1547 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1548 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1549 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1550 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv8bf16(<vscale x 8 x bfloat> [[ZT_COERCE0:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE1:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE2:%.*]], <vscale x 8 x bfloat> [[ZT_COERCE3:%.*]], <vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
1551 // CPP-CHECK-NEXT: ret void
1552 //
1553 void test_svst4q_vnum_bf16(svbool_t pg, const bfloat16_t *base, int64_t vnum, svbfloat16x4_t zt) ATTR
1554 {
1555 SVE_ACLE_FUNC(svst4q_vnum,,_bf16,)(pg, base, vnum, zt);
1556 }
1557
1558 // CHECK-LABEL: @test_svst4q_vnum_f32(
1559 // CHECK-NEXT: entry:
1560 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1561 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1562 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1563 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1564 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1565 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x float> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1566 // CHECK-NEXT: ret void
1567 //
1568 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_f32u10__SVBool_tPKfl13svfloat32x4_t(
1569 // CPP-CHECK-NEXT: entry:
1570 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
1571 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1572 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1573 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1574 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1575 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv4f32(<vscale x 4 x float> [[ZT_COERCE0:%.*]], <vscale x 4 x float> [[ZT_COERCE1:%.*]], <vscale x 4 x float> [[ZT_COERCE2:%.*]], <vscale x 4 x float> [[ZT_COERCE3:%.*]], <vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
1576 // CPP-CHECK-NEXT: ret void
1577 //
1578 void test_svst4q_vnum_f32(svbool_t pg, const float32_t *base, int64_t vnum, svfloat32x4_t zt) ATTR
1579 {
1580 SVE_ACLE_FUNC(svst4q_vnum,,_f32,)(pg, base, vnum, zt);
1581 }
1582
1583 // CHECK-LABEL: @test_svst4q_vnum_f64(
1584 // CHECK-NEXT: entry:
1585 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1586 // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1587 // CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1588 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1589 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1590 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x double> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1591 // CHECK-NEXT: ret void
1592 //
1593 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_f64u10__SVBool_tPKdl13svfloat64x4_t(
1594 // CPP-CHECK-NEXT: entry:
1595 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
1596 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
1597 // CPP-CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
1598 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
1599 // CPP-CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1600 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv2f64(<vscale x 2 x double> [[ZT_COERCE0:%.*]], <vscale x 2 x double> [[ZT_COERCE1:%.*]], <vscale x 2 x double> [[ZT_COERCE2:%.*]], <vscale x 2 x double> [[ZT_COERCE3:%.*]], <vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
1601 // CPP-CHECK-NEXT: ret void
1602 //
1603 void test_svst4q_vnum_f64(svbool_t pg, const float64_t *base, int64_t vnum, svfloat64x4_t zt) ATTR
1604 {
1605 SVE_ACLE_FUNC(svst4q_vnum,,_f64,)(pg, base, vnum, zt);
1606 }
1607
1608 // CHECK-LABEL: @test_svst4q_vnum_mf8(
1609 // CHECK-NEXT: entry:
1610 // CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1611 // CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1612 // CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1613 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1614 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1615 // CHECK-NEXT: ret void
1616 //
1617 // CPP-CHECK-LABEL: @_Z20test_svst4q_vnum_mf8u10__SVBool_tPKu6__mfp8l13svmfloat8x4_t(
1618 // CPP-CHECK-NEXT: entry:
1619 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
1620 // CPP-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
1621 // CPP-CHECK-NEXT: [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
1622 // CPP-CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
1623 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4q.nxv16i8(<vscale x 16 x i8> [[ZT_COERCE0:%.*]], <vscale x 16 x i8> [[ZT_COERCE1:%.*]], <vscale x 16 x i8> [[ZT_COERCE2:%.*]], <vscale x 16 x i8> [[ZT_COERCE3:%.*]], <vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
1624 // CPP-CHECK-NEXT: ret void
1625 //
1626 void test_svst4q_vnum_mf8(svbool_t pg, const mfloat8_t *base, int64_t vnum, svmfloat8x4_t zt) ATTR
1627 {
1628 SVE_ACLE_FUNC(svst4q_vnum,,_mf8,)(pg, base, vnum, zt);
1629 }