clang/test/CodeGen/AArch64/sve2p1-intrinsics/acle_sve2p1_ldnt1.c

   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 +bf16 -target-feature +sme -target-feature +sme2 -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
   4 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
   5 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
   6 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
   7 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
   8 // RUN: %clang_cc1 -triple aarch64 -target-feature +bf16 -target-feature +sme -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
   9
  10 #include <arm_sve.h>
  11
  12 #ifdef SVE_OVERLOADED_FORMS
  13 // A simple used,unused... macro, long enough to represent any SVE builtin.
  14 #define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
  15 #else
  16 #define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
  17 #endif
  18
  19 #ifdef __ARM_FEATURE_SME
  20 #define ATTR __arm_streaming
  21 #else
  22 #define ATTR
  23 #endif
  24
  25 // CHECK-LABEL: @test_svldnt1_u8_x2(
  26 // CHECK-NEXT:  entry:
  27 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  28 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
  29 //
  30 // CPP-CHECK-LABEL: @_Z18test_svldnt1_u8_x2u11__SVCount_tPKh(
  31 // CPP-CHECK-NEXT:  entry:
  32 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  33 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
  34 //
  35 svuint8x2_t test_svldnt1_u8_x2(svcount_t pn, const uint8_t *base) ATTR
  36 {
  37   return SVE_ACLE_FUNC(svldnt1,_u8,_x2,)(pn, base);
  38 }
  39
  40 // CHECK-LABEL: @test_svldnt1_u16_x2(
  41 // CHECK-NEXT:  entry:
  42 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  43 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
  44 //
  45 // CPP-CHECK-LABEL: @_Z19test_svldnt1_u16_x2u11__SVCount_tPKt(
  46 // CPP-CHECK-NEXT:  entry:
  47 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  48 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
  49 //
  50 svuint16x2_t test_svldnt1_u16_x2(svcount_t pn, const uint16_t *base) ATTR
  51 {
  52   return SVE_ACLE_FUNC(svldnt1,_u16,_x2,)(pn, base);
  53 }
  54
  55 // CHECK-LABEL: @test_svldnt1_u32_x2(
  56 // CHECK-NEXT:  entry:
  57 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  58 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
  59 //
  60 // CPP-CHECK-LABEL: @_Z19test_svldnt1_u32_x2u11__SVCount_tPKj(
  61 // CPP-CHECK-NEXT:  entry:
  62 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  63 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
  64 //
  65 svuint32x2_t test_svldnt1_u32_x2(svcount_t pn, const uint32_t *base) ATTR
  66 {
  67   return SVE_ACLE_FUNC(svldnt1,_u32,_x2,)(pn, base);
  68 }
  69
  70 // CHECK-LABEL: @test_svldnt1_u64_x2(
  71 // CHECK-NEXT:  entry:
  72 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  73 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
  74 //
  75 // CPP-CHECK-LABEL: @_Z19test_svldnt1_u64_x2u11__SVCount_tPKm(
  76 // CPP-CHECK-NEXT:  entry:
  77 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  78 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
  79 //
  80 svuint64x2_t test_svldnt1_u64_x2(svcount_t pn, const uint64_t *base) ATTR
  81 {
  82   return SVE_ACLE_FUNC(svldnt1,_u64,_x2,)(pn, base);
  83 }
  84
  85 // CHECK-LABEL: @test_svldnt1_u8_x4(
  86 // CHECK-NEXT:  entry:
  87 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  88 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
  89 //
  90 // CPP-CHECK-LABEL: @_Z18test_svldnt1_u8_x4u11__SVCount_tPKh(
  91 // CPP-CHECK-NEXT:  entry:
  92 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
  93 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
  94 //
  95 svuint8x4_t test_svldnt1_u8_x4(svcount_t pn, const uint8_t *base) ATTR
  96 {
  97   return SVE_ACLE_FUNC(svldnt1,_u8,_x4,)(pn, base);
  98 }
  99
 100 // CHECK-LABEL: @test_svldnt1_u16_x4(
 101 // CHECK-NEXT:  entry:
 102 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 103 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
 104 //
 105 // CPP-CHECK-LABEL: @_Z19test_svldnt1_u16_x4u11__SVCount_tPKt(
 106 // CPP-CHECK-NEXT:  entry:
 107 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 108 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
 109 //
 110 svuint16x4_t test_svldnt1_u16_x4(svcount_t pn, const uint16_t *base) ATTR
 111 {
 112   return SVE_ACLE_FUNC(svldnt1,_u16,_x4,)(pn, base);
 113 }
 114
 115 // CHECK-LABEL: @test_svldnt1_u32_x4(
 116 // CHECK-NEXT:  entry:
 117 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 118 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
 119 //
 120 // CPP-CHECK-LABEL: @_Z19test_svldnt1_u32_x4u11__SVCount_tPKj(
 121 // CPP-CHECK-NEXT:  entry:
 122 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 123 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
 124 //
 125 svuint32x4_t test_svldnt1_u32_x4(svcount_t pn, const uint32_t *base) ATTR
 126 {
 127   return SVE_ACLE_FUNC(svldnt1,_u32,_x4,)(pn, base);
 128 }
 129
 130 // CHECK-LABEL: @test_svldnt1_u64_x4(
 131 // CHECK-NEXT:  entry:
 132 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 133 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
 134 //
 135 // CPP-CHECK-LABEL: @_Z19test_svldnt1_u64_x4u11__SVCount_tPKm(
 136 // CPP-CHECK-NEXT:  entry:
 137 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 138 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
 139 //
 140 svuint64x4_t test_svldnt1_u64_x4(svcount_t pn, const uint64_t *base) ATTR
 141 {
 142   return SVE_ACLE_FUNC(svldnt1,_u64,_x4,)(pn, base);
 143 }
 144
 145 // CHECK-LABEL: @test_svldnt1_s8_x2(
 146 // CHECK-NEXT:  entry:
 147 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 148 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
 149 //
 150 // CPP-CHECK-LABEL: @_Z18test_svldnt1_s8_x2u11__SVCount_tPKa(
 151 // CPP-CHECK-NEXT:  entry:
 152 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 153 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
 154 //
 155 svint8x2_t test_svldnt1_s8_x2(svcount_t pn, const int8_t *base) ATTR
 156 {
 157   return SVE_ACLE_FUNC(svldnt1,_s8,_x2,)(pn, base);
 158 }
 159
 160 // CHECK-LABEL: @test_svldnt1_s16_x2(
 161 // CHECK-NEXT:  entry:
 162 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 163 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
 164 //
 165 // CPP-CHECK-LABEL: @_Z19test_svldnt1_s16_x2u11__SVCount_tPKs(
 166 // CPP-CHECK-NEXT:  entry:
 167 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 168 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
 169 //
 170 svint16x2_t test_svldnt1_s16_x2(svcount_t pn, const int16_t *base) ATTR
 171 {
 172   return SVE_ACLE_FUNC(svldnt1,_s16,_x2,)(pn, base);
 173 }
 174
 175 // CHECK-LABEL: @test_svldnt1_s32_x2(
 176 // CHECK-NEXT:  entry:
 177 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 178 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
 179 //
 180 // CPP-CHECK-LABEL: @_Z19test_svldnt1_s32_x2u11__SVCount_tPKi(
 181 // CPP-CHECK-NEXT:  entry:
 182 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 183 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
 184 //
 185 svint32x2_t test_svldnt1_s32_x2(svcount_t pn, const int32_t *base) ATTR
 186 {
 187   return SVE_ACLE_FUNC(svldnt1,_s32,_x2,)(pn, base);
 188 }
 189
 190 // CHECK-LABEL: @test_svldnt1_s64_x2(
 191 // CHECK-NEXT:  entry:
 192 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 193 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
 194 //
 195 // CPP-CHECK-LABEL: @_Z19test_svldnt1_s64_x2u11__SVCount_tPKl(
 196 // CPP-CHECK-NEXT:  entry:
 197 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 198 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
 199 //
 200 svint64x2_t test_svldnt1_s64_x2(svcount_t pn, const int64_t *base) ATTR
 201 {
 202   return SVE_ACLE_FUNC(svldnt1,_s64,_x2,)(pn, base);
 203 }
 204
 205 // CHECK-LABEL: @test_svldnt1_s8_x4(
 206 // CHECK-NEXT:  entry:
 207 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 208 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
 209 //
 210 // CPP-CHECK-LABEL: @_Z18test_svldnt1_s8_x4u11__SVCount_tPKa(
 211 // CPP-CHECK-NEXT:  entry:
 212 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 213 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
 214 //
 215 svint8x4_t test_svldnt1_s8_x4(svcount_t pn, const int8_t *base) ATTR
 216 {
 217   return SVE_ACLE_FUNC(svldnt1,_s8,_x4,)(pn, base);
 218 }
 219
 220 // CHECK-LABEL: @test_svldnt1_s16_x4(
 221 // CHECK-NEXT:  entry:
 222 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 223 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
 224 //
 225 // CPP-CHECK-LABEL: @_Z19test_svldnt1_s16_x4u11__SVCount_tPKs(
 226 // CPP-CHECK-NEXT:  entry:
 227 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 228 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]]
 229 //
 230 svint16x4_t test_svldnt1_s16_x4(svcount_t pn, const int16_t *base) ATTR
 231 {
 232   return SVE_ACLE_FUNC(svldnt1,_s16,_x4,)(pn, base);
 233 }
 234
 235 // CHECK-LABEL: @test_svldnt1_s32_x4(
 236 // CHECK-NEXT:  entry:
 237 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 238 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
 239 //
 240 // CPP-CHECK-LABEL: @_Z19test_svldnt1_s32_x4u11__SVCount_tPKi(
 241 // CPP-CHECK-NEXT:  entry:
 242 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 243 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]]
 244 //
 245 svint32x4_t test_svldnt1_s32_x4(svcount_t pn, const int32_t *base) ATTR
 246 {
 247   return SVE_ACLE_FUNC(svldnt1,_s32,_x4,)(pn, base);
 248 }
 249
 250 // CHECK-LABEL: @test_svldnt1_s64_x4(
 251 // CHECK-NEXT:  entry:
 252 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 253 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
 254 //
 255 // CPP-CHECK-LABEL: @_Z19test_svldnt1_s64_x4u11__SVCount_tPKl(
 256 // CPP-CHECK-NEXT:  entry:
 257 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 258 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]]
 259 //
 260 svint64x4_t test_svldnt1_s64_x4(svcount_t pn, const int64_t *base) ATTR
 261 {
 262   return SVE_ACLE_FUNC(svldnt1,_s64,_x4,)(pn, base);
 263 }
 264
 265 // CHECK-LABEL: @test_svldnt1_f16_x2(
 266 // CHECK-NEXT:  entry:
 267 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 268 // CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
 269 //
 270 // CPP-CHECK-LABEL: @_Z19test_svldnt1_f16_x2u11__SVCount_tPKDh(
 271 // CPP-CHECK-NEXT:  entry:
 272 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 273 // CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
 274 //
 275 svfloat16x2_t test_svldnt1_f16_x2(svcount_t pn, const float16_t *base) ATTR
 276 {
 277   return SVE_ACLE_FUNC(svldnt1,_f16,_x2,)(pn, base);
 278 }
 279
 280 // CHECK-LABEL: @test_svldnt1_f32_x2(
 281 // CHECK-NEXT:  entry:
 282 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 283 // CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]]
 284 //
 285 // CPP-CHECK-LABEL: @_Z19test_svldnt1_f32_x2u11__SVCount_tPKf(
 286 // CPP-CHECK-NEXT:  entry:
 287 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 288 // CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]]
 289 //
 290 svfloat32x2_t test_svldnt1_f32_x2(svcount_t pn, const float32_t *base) ATTR
 291 {
 292   return SVE_ACLE_FUNC(svldnt1,_f32,_x2,)(pn, base);
 293 }
 294
 295 // CHECK-LABEL: @test_svldnt1_f64_x2(
 296 // CHECK-NEXT:  entry:
 297 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 298 // CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]]
 299 //
 300 // CPP-CHECK-LABEL: @_Z19test_svldnt1_f64_x2u11__SVCount_tPKd(
 301 // CPP-CHECK-NEXT:  entry:
 302 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 303 // CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]]
 304 //
 305 svfloat64x2_t test_svldnt1_f64_x2(svcount_t pn, const float64_t *base) ATTR
 306 {
 307   return SVE_ACLE_FUNC(svldnt1,_f64,_x2,)(pn, base);
 308 }
 309
 310 // CHECK-LABEL: @test_svldnt1_f16_x4(
 311 // CHECK-NEXT:  entry:
 312 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 313 // CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
 314 //
 315 // CPP-CHECK-LABEL: @_Z19test_svldnt1_f16_x4u11__SVCount_tPKDh(
 316 // CPP-CHECK-NEXT:  entry:
 317 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 318 // CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
 319 //
 320 svfloat16x4_t test_svldnt1_f16_x4(svcount_t pn, const float16_t *base) ATTR
 321 {
 322   return SVE_ACLE_FUNC(svldnt1,_f16,_x4,)(pn, base);
 323 }
 324
 325 // CHECK-LABEL: @test_svldnt1_f32_x4(
 326 // CHECK-NEXT:  entry:
 327 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 328 // CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]]
 329 //
 330 // CPP-CHECK-LABEL: @_Z19test_svldnt1_f32_x4u11__SVCount_tPKf(
 331 // CPP-CHECK-NEXT:  entry:
 332 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 333 // CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]]
 334 //
 335 svfloat32x4_t test_svldnt1_f32_x4(svcount_t pn, const float32_t *base) ATTR
 336 {
 337   return SVE_ACLE_FUNC(svldnt1,_f32,_x4,)(pn, base);
 338 }
 339
 340 // CHECK-LABEL: @test_svldnt1_f64_x4(
 341 // CHECK-NEXT:  entry:
 342 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 343 // CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]]
 344 //
 345 // CPP-CHECK-LABEL: @_Z19test_svldnt1_f64_x4u11__SVCount_tPKd(
 346 // CPP-CHECK-NEXT:  entry:
 347 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[BASE:%.*]])
 348 // CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]]
 349 //
 350 svfloat64x4_t test_svldnt1_f64_x4(svcount_t pn, const float64_t *base) ATTR
 351 {
 352   return SVE_ACLE_FUNC(svldnt1,_f64,_x4,)(pn, base);
 353 }
 354
 355
 356 // == VNUM variants ==
 357
 358
 359 // CHECK-LABEL: @test_svldnt1_vnum_u8_x2(
 360 // CHECK-NEXT:  entry:
 361 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 362 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 363 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 364 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 365 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 366 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 367 //
 368 // CPP-CHECK-LABEL: @_Z23test_svldnt1_vnum_u8_x2u11__SVCount_tPKhl(
 369 // CPP-CHECK-NEXT:  entry:
 370 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 371 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 372 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 373 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 374 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 375 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 376 //
 377 svuint8x2_t test_svldnt1_vnum_u8_x2(svcount_t pn, const uint8_t *base, int64_t vnum) ATTR
 378 {
 379   return SVE_ACLE_FUNC(svldnt1_vnum,_u8,_x2,)(pn, base, vnum);
 380 }
 381
 382 // CHECK-LABEL: @test_svldnt1_vnum_u16_x2(
 383 // CHECK-NEXT:  entry:
 384 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 385 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 386 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 387 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 388 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 389 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 390 //
 391 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_u16_x2u11__SVCount_tPKtl(
 392 // CPP-CHECK-NEXT:  entry:
 393 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 394 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 395 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 396 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 397 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 398 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 399 //
 400 svuint16x2_t test_svldnt1_vnum_u16_x2(svcount_t pn, const uint16_t *base, int64_t vnum) ATTR
 401 {
 402   return SVE_ACLE_FUNC(svldnt1_vnum,_u16,_x2,)(pn, base, vnum);
 403 }
 404
 405 // CHECK-LABEL: @test_svldnt1_vnum_u32_x2(
 406 // CHECK-NEXT:  entry:
 407 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 408 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 409 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 410 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 411 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 412 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 413 //
 414 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_u32_x2u11__SVCount_tPKjl(
 415 // CPP-CHECK-NEXT:  entry:
 416 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 417 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 418 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 419 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 420 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 421 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 422 //
 423 svuint32x2_t test_svldnt1_vnum_u32_x2(svcount_t pn, const uint32_t *base, int64_t vnum) ATTR
 424 {
 425   return SVE_ACLE_FUNC(svldnt1_vnum,_u32,_x2,)(pn, base, vnum);
 426 }
 427
 428 // CHECK-LABEL: @test_svldnt1_vnum_u64_x2(
 429 // CHECK-NEXT:  entry:
 430 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 431 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 432 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 433 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 434 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 435 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 436 //
 437 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_u64_x2u11__SVCount_tPKml(
 438 // CPP-CHECK-NEXT:  entry:
 439 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 440 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 441 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 442 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 443 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 444 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 445 //
 446 svuint64x2_t test_svldnt1_vnum_u64_x2(svcount_t pn, const uint64_t *base, int64_t vnum) ATTR
 447 {
 448   return SVE_ACLE_FUNC(svldnt1_vnum,_u64,_x2,)(pn, base, vnum);
 449 }
 450
 451 // CHECK-LABEL: @test_svldnt1_vnum_u8_x4(
 452 // CHECK-NEXT:  entry:
 453 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 454 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 455 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 456 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 457 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 458 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 459 //
 460 // CPP-CHECK-LABEL: @_Z23test_svldnt1_vnum_u8_x4u11__SVCount_tPKhl(
 461 // CPP-CHECK-NEXT:  entry:
 462 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 463 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 464 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 465 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 466 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 467 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 468 //
 469 svuint8x4_t test_svldnt1_vnum_u8_x4(svcount_t pn, const uint8_t *base, int64_t vnum) ATTR
 470 {
 471   return SVE_ACLE_FUNC(svldnt1_vnum,_u8,_x4,)(pn, base, vnum);
 472 }
 473
 474 // CHECK-LABEL: @test_svldnt1_vnum_u16_x4(
 475 // CHECK-NEXT:  entry:
 476 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 477 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 478 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 479 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 480 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 481 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 482 //
 483 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_u16_x4u11__SVCount_tPKtl(
 484 // CPP-CHECK-NEXT:  entry:
 485 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 486 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 487 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 488 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 489 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 490 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 491 //
 492 svuint16x4_t test_svldnt1_vnum_u16_x4(svcount_t pn, const uint16_t *base, int64_t vnum) ATTR
 493 {
 494   return SVE_ACLE_FUNC(svldnt1_vnum,_u16,_x4,)(pn, base, vnum);
 495 }
 496
 497 // CHECK-LABEL: @test_svldnt1_vnum_u32_x4(
 498 // CHECK-NEXT:  entry:
 499 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 500 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 501 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 502 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 503 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 504 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 505 //
 506 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_u32_x4u11__SVCount_tPKjl(
 507 // CPP-CHECK-NEXT:  entry:
 508 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 509 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 510 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 511 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 512 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 513 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 514 //
 515 svuint32x4_t test_svldnt1_vnum_u32_x4(svcount_t pn, const uint32_t *base, int64_t vnum) ATTR
 516 {
 517   return SVE_ACLE_FUNC(svldnt1_vnum,_u32,_x4,)(pn, base, vnum);
 518 }
 519
 520 // CHECK-LABEL: @test_svldnt1_vnum_u64_x4(
 521 // CHECK-NEXT:  entry:
 522 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 523 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 524 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 525 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 526 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 527 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 528 //
 529 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_u64_x4u11__SVCount_tPKml(
 530 // CPP-CHECK-NEXT:  entry:
 531 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 532 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 533 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 534 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 535 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 536 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 537 //
 538 svuint64x4_t test_svldnt1_vnum_u64_x4(svcount_t pn, const uint64_t *base, int64_t vnum) ATTR
 539 {
 540   return SVE_ACLE_FUNC(svldnt1_vnum,_u64,_x4,)(pn, base, vnum);
 541 }
 542
 543 // CHECK-LABEL: @test_svldnt1_vnum_s8_x2(
 544 // CHECK-NEXT:  entry:
 545 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 546 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 547 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 548 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 549 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 550 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 551 //
 552 // CPP-CHECK-LABEL: @_Z23test_svldnt1_vnum_s8_x2u11__SVCount_tPKal(
 553 // CPP-CHECK-NEXT:  entry:
 554 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 555 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 556 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 557 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 558 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 559 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 560 //
 561 svint8x2_t test_svldnt1_vnum_s8_x2(svcount_t pn, const int8_t *base, int64_t vnum) ATTR
 562 {
 563   return SVE_ACLE_FUNC(svldnt1_vnum,_s8,_x2,)(pn, base, vnum);
 564 }
 565
 566 // CHECK-LABEL: @test_svldnt1_vnum_s16_x2(
 567 // CHECK-NEXT:  entry:
 568 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 569 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 570 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 571 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 572 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 573 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 574 //
 575 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_s16_x2u11__SVCount_tPKsl(
 576 // CPP-CHECK-NEXT:  entry:
 577 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 578 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 579 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 580 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 581 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 582 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 583 //
 584 svint16x2_t test_svldnt1_vnum_s16_x2(svcount_t pn, const int16_t *base, int64_t vnum) ATTR
 585 {
 586   return SVE_ACLE_FUNC(svldnt1_vnum,_s16,_x2,)(pn, base, vnum);
 587 }
 588
 589 // CHECK-LABEL: @test_svldnt1_vnum_s32_x2(
 590 // CHECK-NEXT:  entry:
 591 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 592 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 593 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 594 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 595 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 596 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 597 //
 598 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_s32_x2u11__SVCount_tPKil(
 599 // CPP-CHECK-NEXT:  entry:
 600 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 601 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 602 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 603 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 604 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 605 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 606 //
 607 svint32x2_t test_svldnt1_vnum_s32_x2(svcount_t pn, const int32_t *base, int64_t vnum) ATTR
 608 {
 609   return SVE_ACLE_FUNC(svldnt1_vnum,_s32,_x2,)(pn, base, vnum);
 610 }
 611
 612 // CHECK-LABEL: @test_svldnt1_vnum_s64_x2(
 613 // CHECK-NEXT:  entry:
 614 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 615 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 616 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 617 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 618 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 619 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 620 //
 621 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_s64_x2u11__SVCount_tPKll(
 622 // CPP-CHECK-NEXT:  entry:
 623 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 624 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 625 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 626 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 627 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 628 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 629 //
 630 svint64x2_t test_svldnt1_vnum_s64_x2(svcount_t pn, const int64_t *base, int64_t vnum) ATTR
 631 {
 632   return SVE_ACLE_FUNC(svldnt1_vnum,_s64,_x2,)(pn, base, vnum);
 633 }
 634
 635 // CHECK-LABEL: @test_svldnt1_vnum_s8_x4(
 636 // CHECK-NEXT:  entry:
 637 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 638 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 639 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 640 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 641 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 642 // CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 643 //
 644 // CPP-CHECK-LABEL: @_Z23test_svldnt1_vnum_s8_x4u11__SVCount_tPKal(
 645 // CPP-CHECK-NEXT:  entry:
 646 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 647 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 648 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 649 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 650 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv16i8(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 651 // CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
 652 //
 653 svint8x4_t test_svldnt1_vnum_s8_x4(svcount_t pn, const int8_t *base, int64_t vnum) ATTR
 654 {
 655   return SVE_ACLE_FUNC(svldnt1_vnum,_s8,_x4,)(pn, base, vnum);
 656 }
 657
 658 // CHECK-LABEL: @test_svldnt1_vnum_s16_x4(
 659 // CHECK-NEXT:  entry:
 660 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 661 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 662 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 663 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 664 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 665 // CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 666 //
 667 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_s16_x4u11__SVCount_tPKsl(
 668 // CPP-CHECK-NEXT:  entry:
 669 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 670 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 671 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 672 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 673 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8i16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 674 // CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP3]]
 675 //
 676 svint16x4_t test_svldnt1_vnum_s16_x4(svcount_t pn, const int16_t *base, int64_t vnum) ATTR
 677 {
 678   return SVE_ACLE_FUNC(svldnt1_vnum,_s16,_x4,)(pn, base, vnum);
 679 }
 680
 681 // CHECK-LABEL: @test_svldnt1_vnum_s32_x4(
 682 // CHECK-NEXT:  entry:
 683 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 684 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 685 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 686 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 687 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 688 // CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 689 //
 690 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_s32_x4u11__SVCount_tPKil(
 691 // CPP-CHECK-NEXT:  entry:
 692 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 693 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 694 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 695 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 696 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4i32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 697 // CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP3]]
 698 //
 699 svint32x4_t test_svldnt1_vnum_s32_x4(svcount_t pn, const int32_t *base, int64_t vnum) ATTR
 700 {
 701   return SVE_ACLE_FUNC(svldnt1_vnum,_s32,_x4,)(pn, base, vnum);
 702 }
 703
 704 // CHECK-LABEL: @test_svldnt1_vnum_s64_x4(
 705 // CHECK-NEXT:  entry:
 706 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 707 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 708 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 709 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 710 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 711 // CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 712 //
 713 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_s64_x4u11__SVCount_tPKll(
 714 // CPP-CHECK-NEXT:  entry:
 715 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 716 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 717 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 718 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 719 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2i64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 720 // CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP3]]
 721 //
 722 svint64x4_t test_svldnt1_vnum_s64_x4(svcount_t pn, const int64_t *base, int64_t vnum) ATTR
 723 {
 724   return SVE_ACLE_FUNC(svldnt1_vnum,_s64,_x4,)(pn, base, vnum);
 725 }
 726
 727 // CHECK-LABEL: @test_svldnt1_vnum_f16_x2(
 728 // CHECK-NEXT:  entry:
 729 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 730 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 731 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 732 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 733 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 734 // CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP3]]
 735 //
 736 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_f16_x2u11__SVCount_tPKDhl(
 737 // CPP-CHECK-NEXT:  entry:
 738 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 739 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 740 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 741 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 742 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 743 // CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP3]]
 744 //
 745 svfloat16x2_t test_svldnt1_vnum_f16_x2(svcount_t pn, const float16_t *base, int64_t vnum) ATTR
 746 {
 747   return SVE_ACLE_FUNC(svldnt1_vnum,_f16,_x2,)(pn, base, vnum);
 748 }
 749
 750 // CHECK-LABEL: @test_svldnt1_vnum_f32_x2(
 751 // CHECK-NEXT:  entry:
 752 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 753 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 754 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 755 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 756 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 757 // CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP3]]
 758 //
 759 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_f32_x2u11__SVCount_tPKfl(
 760 // CPP-CHECK-NEXT:  entry:
 761 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 762 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 763 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 764 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 765 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 766 // CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP3]]
 767 //
 768 svfloat32x2_t test_svldnt1_vnum_f32_x2(svcount_t pn, const float32_t *base, int64_t vnum) ATTR
 769 {
 770   return SVE_ACLE_FUNC(svldnt1_vnum,_f32,_x2,)(pn, base, vnum);
 771 }
 772
 773 // CHECK-LABEL: @test_svldnt1_vnum_f64_x2(
 774 // CHECK-NEXT:  entry:
 775 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 776 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 777 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 778 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 779 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 780 // CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP3]]
 781 //
 782 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_f64_x2u11__SVCount_tPKdl(
 783 // CPP-CHECK-NEXT:  entry:
 784 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 785 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 786 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 787 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 788 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x2.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 789 // CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP3]]
 790 //
 791 svfloat64x2_t test_svldnt1_vnum_f64_x2(svcount_t pn, const float64_t *base, int64_t vnum) ATTR
 792 {
 793   return SVE_ACLE_FUNC(svldnt1_vnum,_f64,_x2,)(pn, base, vnum);
 794 }
 795
 796 // CHECK-LABEL: @test_svldnt1_vnum_f16_x4(
 797 // CHECK-NEXT:  entry:
 798 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 799 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 800 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 801 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 802 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 803 // CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP3]]
 804 //
 805 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_f16_x4u11__SVCount_tPKDhl(
 806 // CPP-CHECK-NEXT:  entry:
 807 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 808 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 809 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 810 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 811 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv8f16(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 812 // CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP3]]
 813 //
 814 svfloat16x4_t test_svldnt1_vnum_f16_x4(svcount_t pn, const float16_t *base, int64_t vnum) ATTR
 815 {
 816   return SVE_ACLE_FUNC(svldnt1_vnum,_f16,_x4,)(pn, base, vnum);
 817 }
 818
 819 // CHECK-LABEL: @test_svldnt1_vnum_f32_x4(
 820 // CHECK-NEXT:  entry:
 821 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 822 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 823 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 824 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 825 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 826 // CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP3]]
 827 //
 828 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_f32_x4u11__SVCount_tPKfl(
 829 // CPP-CHECK-NEXT:  entry:
 830 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 831 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 832 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 833 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 834 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv4f32(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 835 // CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP3]]
 836 //
 837 svfloat32x4_t test_svldnt1_vnum_f32_x4(svcount_t pn, const float32_t *base, int64_t vnum) ATTR
 838 {
 839   return SVE_ACLE_FUNC(svldnt1_vnum,_f32,_x4,)(pn, base, vnum);
 840 }
 841
 842 // CHECK-LABEL: @test_svldnt1_vnum_f64_x4(
 843 // CHECK-NEXT:  entry:
 844 // CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 845 // CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 846 // CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 847 // CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 848 // CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 849 // CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP3]]
 850 //
 851 // CPP-CHECK-LABEL: @_Z24test_svldnt1_vnum_f64_x4u11__SVCount_tPKdl(
 852 // CPP-CHECK-NEXT:  entry:
 853 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
 854 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl i64 [[VNUM:%.*]], 4
 855 // CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[TMP1]], [[TMP0]]
 856 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
 857 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ldnt1.pn.x4.nxv2f64(target("aarch64.svcount") [[PN:%.*]], ptr [[TMP2]])
 858 // CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP3]]
 859 //
 860 svfloat64x4_t test_svldnt1_vnum_f64_x4(svcount_t pn, const float64_t *base, int64_t vnum) ATTR
 861 {
 862   return SVE_ACLE_FUNC(svldnt1_vnum,_f64,_x4,)(pn, base, vnum);
 863 }