// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -no-opaque-pointers -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -no-opaque-pointers -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -no-opaque-pointers -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -no-opaque-pointers -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -no-opaque-pointers -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -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

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP1]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6:![0-9]+]]
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z16test_svld1sh_s32u10__SVBool_tPKs(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP1]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6:![0-9]+]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
svint32_t test_svld1sh_s32(svbool_t pg, const int16_t *base)
{
  return svld1sh_s32(pg, base);
}

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP1]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z16test_svld1sh_s64u10__SVBool_tPKs(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP1]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
svint64_t test_svld1sh_s64(svbool_t pg, const int16_t *base)
{
  return svld1sh_s64(pg, base);
}

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP1]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z16test_svld1sh_u32u10__SVBool_tPKs(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP1]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
svuint32_t test_svld1sh_u32(svbool_t pg, const int16_t *base)
{
  return svld1sh_u32(pg, base);
}

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP1]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z16test_svld1sh_u64u10__SVBool_tPKs(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP1]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
svuint64_t test_svld1sh_u64(svbool_t pg, const int16_t *base)
{
  return svld1sh_u64(pg, base);
}

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 4 x i16>*
// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP3]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 4 x i16> [[TMP4]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP5]]
//
// CPP-CHECK-LABEL: @_Z21test_svld1sh_vnum_s32u10__SVBool_tPKsl(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 4 x i16>*
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP3]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CPP-CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 4 x i16> [[TMP4]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP5]]
//
svint32_t test_svld1sh_vnum_s32(svbool_t pg, const int16_t *base, int64_t vnum)
{
  return svld1sh_vnum_s32(pg, base, vnum);
}

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 2 x i16>*
// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP3]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 2 x i16> [[TMP4]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP5]]
//
// CPP-CHECK-LABEL: @_Z21test_svld1sh_vnum_s64u10__SVBool_tPKsl(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 2 x i16>*
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP3]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CPP-CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 2 x i16> [[TMP4]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP5]]
//
svint64_t test_svld1sh_vnum_s64(svbool_t pg, const int16_t *base, int64_t vnum)
{
  return svld1sh_vnum_s64(pg, base, vnum);
}

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 4 x i16>*
// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP3]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 4 x i16> [[TMP4]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP5]]
//
// CPP-CHECK-LABEL: @_Z21test_svld1sh_vnum_u32u10__SVBool_tPKsl(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 4 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 4 x i16>*
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i16> @llvm.masked.load.nxv4i16.p0nxv4i16(<vscale x 4 x i16>* [[TMP3]], i32 1, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CPP-CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 4 x i16> [[TMP4]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP5]]
//
svuint32_t test_svld1sh_vnum_u32(svbool_t pg, const int16_t *base, int64_t vnum)
{
  return svld1sh_vnum_u32(pg, base, vnum);
}

// CHECK-LABEL: @test_svld1sh_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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 2 x i16>*
// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP3]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 2 x i16> [[TMP4]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP5]]
//
// CPP-CHECK-LABEL: @_Z21test_svld1sh_vnum_u64u10__SVBool_tPKsl(
// 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:%.*]] = bitcast i16* [[BASE:%.*]] to <vscale x 2 x i16>*
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* [[TMP1]], i64 [[VNUM:%.*]], i64 0
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = bitcast i16* [[TMP2]] to <vscale x 2 x i16>*
// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x i16> @llvm.masked.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* [[TMP3]], i32 1, <vscale x 2 x i1> [[TMP0]], <vscale x 2 x i16> zeroinitializer), !tbaa [[TBAA6]]
// CPP-CHECK-NEXT:    [[TMP5:%.*]] = sext <vscale x 2 x i16> [[TMP4]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP5]]
//
svuint64_t test_svld1sh_vnum_u64(svbool_t pg, const int16_t *base, int64_t vnum)
{
  return svld1sh_vnum_u64(pg, base, vnum);
}

// CHECK-LABEL: @test_svld1sh_gather_u32base_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 0)
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z31test_svld1sh_gather_u32base_s32u10__SVBool_tu12__SVUint32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 0)
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svint32_t test_svld1sh_gather_u32base_s32(svbool_t pg, svuint32_t bases) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u32base, _s32, )(pg, bases);
}

// CHECK-LABEL: @test_svld1sh_gather_u64base_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 0)
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z31test_svld1sh_gather_u64base_s64u10__SVBool_tu12__SVUint64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 0)
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svint64_t test_svld1sh_gather_u64base_s64(svbool_t pg, svuint64_t bases) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u64base, _s64, )(pg, bases);
}

// CHECK-LABEL: @test_svld1sh_gather_u32base_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 0)
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z31test_svld1sh_gather_u32base_u32u10__SVBool_tu12__SVUint32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 0)
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svuint32_t test_svld1sh_gather_u32base_u32(svbool_t pg, svuint32_t bases) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u32base, _u32, )(pg, bases);
}

// CHECK-LABEL: @test_svld1sh_gather_u64base_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 0)
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z31test_svld1sh_gather_u64base_u64u10__SVBool_tu12__SVUint64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 0)
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svuint64_t test_svld1sh_gather_u64base_u64(svbool_t pg, svuint64_t bases) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u64base, _u64, )(pg, bases);
}

// CHECK-LABEL: @test_svld1sh_gather_s32offset_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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_s32offset_s32u10__SVBool_tPKsu11__SVInt32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svint32_t test_svld1sh_gather_s32offset_s32(svbool_t pg, const int16_t *base, svint32_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s32, offset_s32, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_s64offset_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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_s64offset_s64u10__SVBool_tPKsu11__SVInt64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svint64_t test_svld1sh_gather_s64offset_s64(svbool_t pg, const int16_t *base, svint64_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s64, offset_s64, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_s32offset_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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_s32offset_u32u10__SVBool_tPKsu11__SVInt32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svuint32_t test_svld1sh_gather_s32offset_u32(svbool_t pg, const int16_t *base, svint32_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s32, offset_u32, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_s64offset_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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_s64offset_u64u10__SVBool_tPKsu11__SVInt64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svuint64_t test_svld1sh_gather_s64offset_u64(svbool_t pg, const int16_t *base, svint64_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s64, offset_u64, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_u32offset_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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_u32offset_s32u10__SVBool_tPKsu12__SVUint32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svint32_t test_svld1sh_gather_u32offset_s32(svbool_t pg, const int16_t *base, svuint32_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u32, offset_s32, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_u64offset_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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_u64offset_s64u10__SVBool_tPKsu12__SVUint64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svint64_t test_svld1sh_gather_u64offset_s64(svbool_t pg, const int16_t *base, svuint64_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u64, offset_s64, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_u32offset_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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_u32offset_u32u10__SVBool_tPKsu12__SVUint32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svuint32_t test_svld1sh_gather_u32offset_u32(svbool_t pg, const int16_t *base, svuint32_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u32, offset_u32, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_u64offset_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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z33test_svld1sh_gather_u64offset_u64u10__SVBool_tPKsu12__SVUint64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[OFFSETS:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svuint64_t test_svld1sh_gather_u64offset_u64(svbool_t pg, const int16_t *base, svuint64_t offsets) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u64, offset_u64, )(pg, base, offsets);
}

// CHECK-LABEL: @test_svld1sh_gather_u32base_offset_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z38test_svld1sh_gather_u32base_offset_s32u10__SVBool_tu12__SVUint32_tl(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svint32_t test_svld1sh_gather_u32base_offset_s32(svbool_t pg, svuint32_t bases, int64_t offset) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u32base, _offset_s32, )(pg, bases, offset);
}

// CHECK-LABEL: @test_svld1sh_gather_u64base_offset_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z38test_svld1sh_gather_u64base_offset_s64u10__SVBool_tu12__SVUint64_tl(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svint64_t test_svld1sh_gather_u64base_offset_s64(svbool_t pg, svuint64_t bases, int64_t offset) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u64base, _offset_s64, )(pg, bases, offset);
}

// CHECK-LABEL: @test_svld1sh_gather_u32base_offset_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z38test_svld1sh_gather_u32base_offset_u32u10__SVBool_tu12__SVUint32_tl(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svuint32_t test_svld1sh_gather_u32base_offset_u32(svbool_t pg, svuint32_t bases, int64_t offset) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u32base, _offset_u32, )(pg, bases, offset);
}

// CHECK-LABEL: @test_svld1sh_gather_u64base_offset_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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z38test_svld1sh_gather_u64base_offset_u64u10__SVBool_tu12__SVUint64_tl(
// 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 i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[OFFSET:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svuint64_t test_svld1sh_gather_u64base_offset_u64(svbool_t pg, svuint64_t bases, int64_t offset) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u64base, _offset_u64, )(pg, bases, offset);
}

// CHECK-LABEL: @test_svld1sh_gather_s32index_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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_s32index_s32u10__SVBool_tPKsu11__SVInt32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svint32_t test_svld1sh_gather_s32index_s32(svbool_t pg, const int16_t *base, svint32_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s32, index_s32, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_s64index_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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_s64index_s64u10__SVBool_tPKsu11__SVInt64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svint64_t test_svld1sh_gather_s64index_s64(svbool_t pg, const int16_t *base, svint64_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s64, index_s64, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_s32index_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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_s32index_u32u10__SVBool_tPKsu11__SVInt32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.sxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svuint32_t test_svld1sh_gather_s32index_u32(svbool_t pg, const int16_t *base, svint32_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s32, index_u32, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_s64index_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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_s64index_u64u10__SVBool_tPKsu11__SVInt64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svuint64_t test_svld1sh_gather_s64index_u64(svbool_t pg, const int16_t *base, svint64_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, s64, index_u64, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_u32index_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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_u32index_s32u10__SVBool_tPKsu12__SVUint32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svint32_t test_svld1sh_gather_u32index_s32(svbool_t pg, const int16_t *base, svuint32_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u32, index_s32, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_u64index_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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_u64index_s64u10__SVBool_tPKsu12__SVUint64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svint64_t test_svld1sh_gather_u64index_s64(svbool_t pg, const int16_t *base, svuint64_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u64, index_s64, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_u32index_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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_u32index_u32u10__SVBool_tPKsu12__SVUint32_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.uxtw.index.nxv4i16(<vscale x 4 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 4 x i32> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 4 x i16> [[TMP1]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
//
svuint32_t test_svld1sh_gather_u32index_u32(svbool_t pg, const int16_t *base, svuint32_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u32, index_u32, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_u64index_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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z32test_svld1sh_gather_u64index_u64u10__SVBool_tPKsu12__SVUint64_t(
// 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 i16> @llvm.aarch64.sve.ld1.gather.index.nxv2i16(<vscale x 2 x i1> [[TMP0]], i16* [[BASE:%.*]], <vscale x 2 x i64> [[INDICES:%.*]])
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = sext <vscale x 2 x i16> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
//
svuint64_t test_svld1sh_gather_u64index_u64(svbool_t pg, const int16_t *base, svuint64_t indices) {
  return SVE_ACLE_FUNC(svld1sh_gather_, u64, index_u64, )(pg, base, indices);
}

// CHECK-LABEL: @test_svld1sh_gather_u32base_index_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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z37test_svld1sh_gather_u32base_index_s32u10__SVBool_tu12__SVUint32_tl(
// 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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[TMP1]])
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
svint32_t test_svld1sh_gather_u32base_index_s32(svbool_t pg, svuint32_t bases, int64_t index) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u32base, _index_s32, )(pg, bases, index);
}

// CHECK-LABEL: @test_svld1sh_gather_u64base_index_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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z37test_svld1sh_gather_u64base_index_s64u10__SVBool_tu12__SVUint64_tl(
// 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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[TMP1]])
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
svint64_t test_svld1sh_gather_u64base_index_s64(svbool_t pg, svuint64_t bases, int64_t index) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u64base, _index_s64, )(pg, bases, index);
}

// CHECK-LABEL: @test_svld1sh_gather_u32base_index_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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z37test_svld1sh_gather_u32base_index_u32u10__SVBool_tu12__SVUint32_tl(
// 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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv4i16.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[BASES:%.*]], i64 [[TMP1]])
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 4 x i16> [[TMP2]] to <vscale x 4 x i32>
// CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
//
svuint32_t test_svld1sh_gather_u32base_index_u32(svbool_t pg, svuint32_t bases, int64_t index) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u32base, _index_u32, )(pg, bases, index);
}

// CHECK-LABEL: @test_svld1sh_gather_u64base_index_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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z37test_svld1sh_gather_u64base_index_u64u10__SVBool_tu12__SVUint64_tl(
// 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:%.*]] = shl i64 [[INDEX:%.*]], 1
// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i16> @llvm.aarch64.sve.ld1.gather.scalar.offset.nxv2i16.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[BASES:%.*]], i64 [[TMP1]])
// CPP-CHECK-NEXT:    [[TMP3:%.*]] = sext <vscale x 2 x i16> [[TMP2]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
//
svuint64_t test_svld1sh_gather_u64base_index_u64(svbool_t pg, svuint64_t bases, int64_t index) {
  return SVE_ACLE_FUNC(svld1sh_gather, _u64base, _index_u64, )(pg, bases, index);
}