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 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
4 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
5 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -S -disable-O0-optnone -Werror -o /dev/null %s
8 // CHECK-LABEL: @test_svldnf1sh_s32(
10 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
11 // CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
12 // CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
13 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
15 // CPP-CHECK-LABEL: @_Z18test_svldnf1sh_s32u10__SVBool_tPKs(
16 // CPP-CHECK-NEXT: entry:
17 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
18 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
19 // CPP-CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
20 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
22 svint32_t
test_svldnf1sh_s32(svbool_t pg
, const int16_t *base
)
24 return svldnf1sh_s32(pg
, base
);
27 // CHECK-LABEL: @test_svldnf1sh_s64(
29 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
30 // CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
31 // CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
32 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
34 // CPP-CHECK-LABEL: @_Z18test_svldnf1sh_s64u10__SVBool_tPKs(
35 // CPP-CHECK-NEXT: entry:
36 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
37 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
38 // CPP-CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
39 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
41 svint64_t
test_svldnf1sh_s64(svbool_t pg
, const int16_t *base
)
43 return svldnf1sh_s64(pg
, base
);
46 // CHECK-LABEL: @test_svldnf1sh_u32(
48 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
49 // CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
50 // CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
51 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
53 // CPP-CHECK-LABEL: @_Z18test_svldnf1sh_u32u10__SVBool_tPKs(
54 // CPP-CHECK-NEXT: entry:
55 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
56 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
57 // CPP-CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
58 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
60 svuint32_t
test_svldnf1sh_u32(svbool_t pg
, const int16_t *base
)
62 return svldnf1sh_u32(pg
, base
);
65 // CHECK-LABEL: @test_svldnf1sh_u64(
67 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
68 // CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
69 // CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
70 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
72 // CPP-CHECK-LABEL: @_Z18test_svldnf1sh_u64u10__SVBool_tPKs(
73 // CPP-CHECK-NEXT: entry:
74 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
75 // CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
76 // CPP-CHECK-NEXT: [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
77 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
79 svuint64_t
test_svldnf1sh_u64(svbool_t pg
, const int16_t *base
)
81 return svldnf1sh_u64(pg
, base
);
84 // CHECK-LABEL: @test_svldnf1sh_vnum_s32(
86 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
87 // CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 4 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
88 // CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[TMP1]])
89 // CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
90 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
92 // CPP-CHECK-LABEL: @_Z23test_svldnf1sh_vnum_s32u10__SVBool_tPKsl(
93 // CPP-CHECK-NEXT: entry:
94 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
95 // CPP-CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 4 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
96 // CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[TMP1]])
97 // CPP-CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
98 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
100 svint32_t
test_svldnf1sh_vnum_s32(svbool_t pg
, const int16_t *base
, int64_t vnum
)
102 return svldnf1sh_vnum_s32(pg
, base
, vnum
);
105 // CHECK-LABEL: @test_svldnf1sh_vnum_s64(
106 // CHECK-NEXT: entry:
107 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
108 // CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
109 // CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[TMP1]])
110 // CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
111 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
113 // CPP-CHECK-LABEL: @_Z23test_svldnf1sh_vnum_s64u10__SVBool_tPKsl(
114 // CPP-CHECK-NEXT: entry:
115 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
116 // CPP-CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
117 // CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[TMP1]])
118 // CPP-CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
119 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
121 svint64_t
test_svldnf1sh_vnum_s64(svbool_t pg
, const int16_t *base
, int64_t vnum
)
123 return svldnf1sh_vnum_s64(pg
, base
, vnum
);
126 // CHECK-LABEL: @test_svldnf1sh_vnum_u32(
127 // CHECK-NEXT: entry:
128 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
129 // CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 4 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
130 // CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[TMP1]])
131 // CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
132 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
134 // CPP-CHECK-LABEL: @_Z23test_svldnf1sh_vnum_u32u10__SVBool_tPKsl(
135 // CPP-CHECK-NEXT: entry:
136 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
137 // CPP-CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 4 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
138 // CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> [[TMP0]], ptr [[TMP1]])
139 // CPP-CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
140 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
142 svuint32_t
test_svldnf1sh_vnum_u32(svbool_t pg
, const int16_t *base
, int64_t vnum
)
144 return svldnf1sh_vnum_u32(pg
, base
, vnum
);
147 // CHECK-LABEL: @test_svldnf1sh_vnum_u64(
148 // CHECK-NEXT: entry:
149 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
150 // CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
151 // CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[TMP1]])
152 // CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
153 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
155 // CPP-CHECK-LABEL: @_Z23test_svldnf1sh_vnum_u64u10__SVBool_tPKsl(
156 // CPP-CHECK-NEXT: entry:
157 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
158 // CPP-CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i16>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
159 // CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> [[TMP0]], ptr [[TMP1]])
160 // CPP-CHECK-NEXT: [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
161 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
163 svuint64_t
test_svldnf1sh_vnum_u64(svbool_t pg
, const int16_t *base
, int64_t vnum
)
165 return svldnf1sh_vnum_u64(pg
, base
, vnum
);