// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// REQUIRES: aarch64-registered-target
// 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
// 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
// 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
// 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
// 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
// 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
// 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
// RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +sve2 -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s

#include <arm_sve.h>

#ifdef SVE_OVERLOADED_FORMS
// A simple used,unused... macro, long enough to represent any SVE builtin.
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
#else
#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
#endif

#if defined(__ARM_FEATURE_SME) && defined(__ARM_FEATURE_SVE)
#define ATTR __arm_streaming_compatible
#elif defined(__ARM_FEATURE_SME)
#define ATTR __arm_streaming
#else
#define ATTR
#endif

// CHECK-LABEL: @test_svld2q_u8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z14test_svld2q_u8u10__SVBool_tPKh(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
svuint8x2_t test_svld2q_u8(svbool_t pg, const uint8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_u8,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_s8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z14test_svld2q_s8u10__SVBool_tPKa(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
svint8x2_t test_svld2q_s8(svbool_t pg, const int8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_s8,)(pg, base);
}
// CHECK-LABEL: @test_svld2q_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_u16u10__SVBool_tPKt(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
svuint16x2_t test_svld2q_u16(svbool_t pg, const uint16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_u16,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_s16u10__SVBool_tPKs(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
svint16x2_t test_svld2q_s16(svbool_t pg, const int16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_s16,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_u32u10__SVBool_tPKj(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
svuint32x2_t test_svld2q_u32(svbool_t pg, const uint32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_u32,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_s32u10__SVBool_tPKi(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
svint32x2_t test_svld2q_s32(svbool_t pg, const int32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_s32,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_u64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_u64u10__SVBool_tPKm(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
svuint64x2_t test_svld2q_u64(svbool_t pg, const uint64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_u64,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_s64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_s64u10__SVBool_tPKl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
svint64x2_t test_svld2q_s64(svbool_t pg, const int64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_s64,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_f16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld2q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_f16u10__SVBool_tPKDh(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld2q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP1]]
//
svfloat16x2_t test_svld2q_f16(svbool_t pg, const float16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_f16,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_bf16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld2q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svld2q_bf16u10__SVBool_tPKu6__bf16(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld2q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]]
//
svbfloat16x2_t test_svld2q_bf16(svbool_t pg, const bfloat16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_bf16,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_f32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld2q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_f32u10__SVBool_tPKf(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld2q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP1]]
//
svfloat32x2_t test_svld2q_f32(svbool_t pg, const float32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_f32,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_f64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld2q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_f64u10__SVBool_tPKd(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld2q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP1]]
//
svfloat64x2_t test_svld2q_f64(svbool_t pg, const float64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_f64,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_mf8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svld2q_mf8u10__SVBool_tPKu6__mfp8(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
svmfloat8x2_t test_svld2q_mf8(svbool_t pg, const mfloat8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld2q,,_mf8,)(pg, base);
}

// CHECK-LABEL: @test_svld2q_vnum_u8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z19test_svld2q_vnum_u8u10__SVBool_tPKhl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
svuint8x2_t test_svld2q_vnum_u8(svbool_t pg, const uint8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum_,,u8,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_s8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z19test_svld2q_vnum_s8u10__SVBool_tPKal(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
svint8x2_t test_svld2q_vnum_s8(svbool_t pg, const int8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_s8,)(pg, base, vnum);
}
// CHECK-LABEL: @test_svld2q_vnum_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_u16u10__SVBool_tPKtl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
svuint16x2_t test_svld2q_vnum_u16(svbool_t pg, const uint16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_u16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_s16u10__SVBool_tPKsl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld2q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
svint16x2_t test_svld2q_vnum_s16(svbool_t pg, const int16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_s16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_u32u10__SVBool_tPKjl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
svuint32x2_t test_svld2q_vnum_u32(svbool_t pg, const uint32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_u32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_s32u10__SVBool_tPKil(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld2q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
svint32x2_t test_svld2q_vnum_s32(svbool_t pg, const int32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_s32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_u64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_u64u10__SVBool_tPKml(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
svuint64x2_t test_svld2q_vnum_u64(svbool_t pg, const uint64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_u64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_s64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_s64u10__SVBool_tPKll(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld2q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
svint64x2_t test_svld2q_vnum_s64(svbool_t pg, const int64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_s64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_f16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld2q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_f16u10__SVBool_tPKDhl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld2q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP4]]
//
svfloat16x2_t test_svld2q_vnum_f16(svbool_t pg, const float16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_f16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_bf16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld2q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z21test_svld2q_vnum_bf16u10__SVBool_tPKu6__bf16l(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld2q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP4]]
//
svbfloat16x2_t test_svld2q_vnum_bf16(svbool_t pg, const bfloat16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_bf16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_f32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld2q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_f32u10__SVBool_tPKfl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld2q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP4]]
//
svfloat32x2_t test_svld2q_vnum_f32(svbool_t pg, const float32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_f32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_f64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld2q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_f64u10__SVBool_tPKdl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld2q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP4]]
//
svfloat64x2_t test_svld2q_vnum_f64(svbool_t pg, const float64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_f64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld2q_vnum_mf8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z20test_svld2q_vnum_mf8u10__SVBool_tPKu6__mfp8l(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld2q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
svmfloat8x2_t test_svld2q_vnum_mf8(svbool_t pg, const mfloat8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld2q_vnum,,_mf8,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_u8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z14test_svld3q_u8u10__SVBool_tPKh(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
svuint8x3_t test_svld3q_u8(svbool_t pg, const uint8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_u8,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_s8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z14test_svld3q_s8u10__SVBool_tPKa(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
svint8x3_t test_svld3q_s8(svbool_t pg, const int8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_s8,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_u16u10__SVBool_tPKt(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
svuint16x3_t test_svld3q_u16(svbool_t pg, const uint16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_u16,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_s16u10__SVBool_tPKs(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
svint16x3_t test_svld3q_s16(svbool_t pg, const int16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_s16,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_u32u10__SVBool_tPKj(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
svuint32x3_t test_svld3q_u32(svbool_t pg, const uint32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_u32,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_s32u10__SVBool_tPKi(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
svint32x3_t test_svld3q_s32(svbool_t pg, const int32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_s32,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_u64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_u64u10__SVBool_tPKm(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
svuint64x3_t test_svld3q_u64(svbool_t pg, const uint64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_u64,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_s64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_s64u10__SVBool_tPKl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
svint64x3_t test_svld3q_s64(svbool_t pg, const int64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_s64,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_f16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld3q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_f16u10__SVBool_tPKDh(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld3q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP1]]
//
svfloat16x3_t test_svld3q_f16(svbool_t pg, const float16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_f16,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_bf16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld3q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svld3q_bf16u10__SVBool_tPKu6__bf16(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld3q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]]
//
svbfloat16x3_t test_svld3q_bf16(svbool_t pg, const bfloat16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_bf16,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_f32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld3q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_f32u10__SVBool_tPKf(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld3q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP1]]
//
svfloat32x3_t test_svld3q_f32(svbool_t pg, const float32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_f32,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_f64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld3q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_f64u10__SVBool_tPKd(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld3q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP1]]
//
svfloat64x3_t test_svld3q_f64(svbool_t pg, const float64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_f64,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_mf8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svld3q_mf8u10__SVBool_tPKu6__mfp8(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
svmfloat8x3_t test_svld3q_mf8(svbool_t pg, const mfloat8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld3q,,_mf8,)(pg, base);
}

// CHECK-LABEL: @test_svld3q_vnum_u8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z19test_svld3q_vnum_u8u10__SVBool_tPKhl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
svuint8x3_t test_svld3q_vnum_u8(svbool_t pg, const uint8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum_,,u8,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_s8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z19test_svld3q_vnum_s8u10__SVBool_tPKal(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
svint8x3_t test_svld3q_vnum_s8(svbool_t pg, const int8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_s8,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_u16u10__SVBool_tPKtl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
svuint16x3_t test_svld3q_vnum_u16(svbool_t pg, const uint16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_u16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_s16u10__SVBool_tPKsl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld3q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
svint16x3_t test_svld3q_vnum_s16(svbool_t pg, const int16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_s16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_u32u10__SVBool_tPKjl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
svuint32x3_t test_svld3q_vnum_u32(svbool_t pg, const uint32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_u32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_s32u10__SVBool_tPKil(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.ld3q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
svint32x3_t test_svld3q_vnum_s32(svbool_t pg, const int32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_s32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_u64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_u64u10__SVBool_tPKml(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
svuint64x3_t test_svld3q_vnum_u64(svbool_t pg, const uint64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_u64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_s64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_s64u10__SVBool_tPKll(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
svint64x3_t test_svld3q_vnum_s64(svbool_t pg, const int64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_s64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_f16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld3q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_f16u10__SVBool_tPKDhl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.ld3q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP4]]
//
svfloat16x3_t test_svld3q_vnum_f16(svbool_t pg, const float16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_f16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_bf16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld3q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z21test_svld3q_vnum_bf16u10__SVBool_tPKu6__bf16l(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld3q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP4]]
//
svbfloat16x3_t test_svld3q_vnum_bf16(svbool_t pg, const bfloat16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_bf16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_f32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld3q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_f32u10__SVBool_tPKfl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld3q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP4]]
//
svfloat32x3_t test_svld3q_vnum_f32(svbool_t pg, const float32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_f32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_f64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld3q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_f64u10__SVBool_tPKdl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld3q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP4]]
//
svfloat64x3_t test_svld3q_vnum_f64(svbool_t pg, const float64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_f64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld3q_vnum_mf8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z20test_svld3q_vnum_mf8u10__SVBool_tPKu6__mfp8l(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld3q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CPP-CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
svmfloat8x3_t test_svld3q_vnum_mf8(svbool_t pg, const mfloat8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld3q_vnum,,_mf8,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_u8(
// CHECK-NEXT:  entry:
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z14test_svld4q_u8u10__SVBool_tPKh(
// CPP-CHECK-NEXT:  entry:
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// 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]]
//
svuint8x4_t test_svld4q_u8(svbool_t pg, const uint8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_u8,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_u16u10__SVBool_tPKt(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
svuint16x4_t test_svld4q_u16(svbool_t pg, const uint16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_u16,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_s16u10__SVBool_tPKs(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP1]]
//
svint16x4_t test_svld4q_s16(svbool_t pg, const int16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_s16,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_u32u10__SVBool_tPKj(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
svuint32x4_t test_svld4q_u32(svbool_t pg, const uint32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_u32,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_s32u10__SVBool_tPKi(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP1]]
//
svint32x4_t test_svld4q_s32(svbool_t pg, const int32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_s32,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_u64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_u64u10__SVBool_tPKm(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
svuint64x4_t test_svld4q_u64(svbool_t pg, const uint64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_u64,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_s64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_s64u10__SVBool_tPKl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP1]]
//
svint64x4_t test_svld4q_s64(svbool_t pg, const int64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_s64,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_f16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_f16u10__SVBool_tPKDh(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP1]]
//
svfloat16x4_t test_svld4q_f16(svbool_t pg, const float16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_f16,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_bf16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld4q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svld4q_bf16u10__SVBool_tPKu6__bf16(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld4q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]]
//
svbfloat16x4_t test_svld4q_bf16(svbool_t pg, const bfloat16_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_bf16,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_f32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_f32u10__SVBool_tPKf(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP1]]
//
svfloat32x4_t test_svld4q_f32(svbool_t pg, const float32_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_f32,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_f64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_f64u10__SVBool_tPKd(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = 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.ld4q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[BASE:%.*]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP1]]
//
svfloat64x4_t test_svld4q_f64(svbool_t pg, const float64_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_f64,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_mf8(
// CHECK-NEXT:  entry:
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svld4q_mf8u10__SVBool_tPKu6__mfp8(
// CPP-CHECK-NEXT:  entry:
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[BASE:%.*]])
// 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]]
//
svmfloat8x4_t test_svld4q_mf8(svbool_t pg, const mfloat8_t *base) ATTR
{
  return SVE_ACLE_FUNC(svld4q,,_mf8,)(pg, base);
}

// CHECK-LABEL: @test_svld4q_vnum_u8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z19test_svld4q_vnum_u8u10__SVBool_tPKhl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// 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]]
//
svuint8x4_t test_svld4q_vnum_u8(svbool_t pg, const uint8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum_,,u8,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_s8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z19test_svld4q_vnum_s8u10__SVBool_tPKal(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// 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]]
//
svint8x4_t test_svld4q_vnum_s8(svbool_t pg, const int8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_s8,)(pg, base, vnum);
}
// CHECK-LABEL: @test_svld4q_vnum_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_u16u10__SVBool_tPKtl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
svuint16x4_t test_svld4q_vnum_u16(svbool_t pg, const uint16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_u16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_s16u10__SVBool_tPKsl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv8i16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP4]]
//
svint16x4_t test_svld4q_vnum_s16(svbool_t pg, const int16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_s16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_u32u10__SVBool_tPKjl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
svuint32x4_t test_svld4q_vnum_u32(svbool_t pg, const uint32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_u32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_s32u10__SVBool_tPKil(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv4i32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP4]]
//
svint32x4_t test_svld4q_vnum_s32(svbool_t pg, const int32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_s32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_u64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_u64u10__SVBool_tPKml(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
svuint64x4_t test_svld4q_vnum_u64(svbool_t pg, const uint64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_u64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_s64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_s64u10__SVBool_tPKll(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv2i64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP4]]
//
svint64x4_t test_svld4q_vnum_s64(svbool_t pg, const int64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_s64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_f16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_f16u10__SVBool_tPKDhl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv8f16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP4]]
//
svfloat16x4_t test_svld4q_vnum_f16(svbool_t pg, const float16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_f16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_bf16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld4q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z21test_svld4q_vnum_bf16u10__SVBool_tPKu6__bf16l(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.ld4q.sret.nxv8bf16(<vscale x 8 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP4]]
//
svbfloat16x4_t test_svld4q_vnum_bf16(svbool_t pg, const bfloat16_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_bf16,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_f32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_f32u10__SVBool_tPKfl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv4f32(<vscale x 4 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP4]]
//
svfloat32x4_t test_svld4q_vnum_f32(svbool_t pg, const float32_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_f32,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_f64(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP4]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_f64u10__SVBool_tPKdl(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP2]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = 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.ld4q.sret.nxv2f64(<vscale x 2 x i1> [[TMP0]], ptr [[TMP3]])
// CPP-CHECK-NEXT:    ret { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP4]]
//
svfloat64x4_t test_svld4q_vnum_f64(svbool_t pg, const float64_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_f64,)(pg, base, vnum);
}

// CHECK-LABEL: @test_svld4q_vnum_mf8(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// CHECK-NEXT:    ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z20test_svld4q_vnum_mf8u10__SVBool_tPKu6__mfp8l(
// CPP-CHECK-NEXT:  entry:
// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call i64 @llvm.vscale.i64()
// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 4
// CPP-CHECK-NEXT:    [[DOTIDX:%.*]] = mul i64 [[VNUM:%.*]], [[TMP1]]
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[DOTIDX]]
// 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.ld4q.sret.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], ptr [[TMP2]])
// 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]]
//
svmfloat8x4_t test_svld4q_vnum_mf8(svbool_t pg, const mfloat8_t *base, int64_t vnum) ATTR
{
  return SVE_ACLE_FUNC(svld4q_vnum,,_mf8,)(pg, base, vnum);
}