// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// RUN: %clang_cc1 -triple thumbv8.1m.main-none-none-eabi -target-feature +mve.fp -mfloat-abi hard -fallow-half-arguments-and-returns -O0 -disable-O0-optnone -S -emit-llvm -o - %s | opt -S -mem2reg -sroa | FileCheck %s
// RUN: %clang_cc1 -triple thumbv8.1m.main-none-none-eabi -target-feature +mve.fp -mfloat-abi hard -fallow-half-arguments-and-returns -O0 -disable-O0-optnone -DPOLYMORPHIC -S -emit-llvm -o - %s | opt -S -mem2reg -sroa | FileCheck %s

// REQUIRES: aarch64-registered-target || arm-registered-target

#include <arm_mve.h>

// CHECK-LABEL: @test_vmlaldavaq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlaldavaq_s16(int64_t a, int16x8_t b, int16x8_t c) {
#ifdef POLYMORPHIC
  return vmlaldavaq(a, b, c);
#else
  return vmlaldavaq_s16(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlaldavaq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlaldavaq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vmlaldavaq(a, b, c);
#else
  return vmlaldavaq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlaldavaq_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 1, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
uint64_t test_vmlaldavaq_u16(uint64_t a, uint16x8_t b, uint16x8_t c) {
#ifdef POLYMORPHIC
  return vmlaldavaq(a, b, c);
#else
  return vmlaldavaq_u16(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlaldavaq_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 1, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
uint64_t test_vmlaldavaq_u32(uint64_t a, uint32x4_t b, uint32x4_t c) {
#ifdef POLYMORPHIC
  return vmlaldavaq(a, b, c);
#else
  return vmlaldavaq_u32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlaldavaxq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 0, i32 1, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlaldavaxq_s16(int64_t a, int16x8_t b, int16x8_t c) {
#ifdef POLYMORPHIC
  return vmlaldavaxq(a, b, c);
#else
  return vmlaldavaxq_s16(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlaldavaxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 0, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlaldavaxq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vmlaldavaxq(a, b, c);
#else
  return vmlaldavaxq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlsldavaq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 1, i32 0, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlsldavaq_s16(int64_t a, int16x8_t b, int16x8_t c) {
#ifdef POLYMORPHIC
  return vmlsldavaq(a, b, c);
#else
  return vmlsldavaq_s16(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlsldavaq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 1, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlsldavaq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vmlsldavaq(a, b, c);
#else
  return vmlsldavaq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlsldaxvaq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 1, i32 1, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlsldaxvaq_s16(int64_t a, int16x8_t b, int16x8_t c) {
#ifdef POLYMORPHIC
  return vmlsldavaxq(a, b, c);
#else
  return vmlsldavaxq_s16(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlsldavaxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 1, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vmlsldavaxq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vmlsldavaxq(a, b, c);
#else
  return vmlsldavaxq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhaq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vrmlaldavhaq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vrmlaldavhaq(a, b, c);
#else
  return vrmlaldavhaq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhaq_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 1, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
uint64_t test_vrmlaldavhaq_u32(uint64_t a, uint32x4_t b, uint32x4_t c) {
#ifdef POLYMORPHIC
  return vrmlaldavhaq(a, b, c);
#else
  return vrmlaldavhaq_u32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhaxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 0, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vrmlaldavhaxq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vrmlaldavhaxq(a, b, c);
#else
  return vrmlaldavhaxq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhaq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 1, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vrmlsldavhaq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vrmlsldavhaq(a, b, c);
#else
  return vrmlsldavhaq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhaxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 1, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]])
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP3]], 1
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = shl i64 [[TMP5]], 32
// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { i32, i32 } [[TMP3]], 0
// CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP6]], [[TMP8]]
// CHECK-NEXT:    ret i64 [[TMP9]]
//
int64_t test_vrmlsldavhaxq_s32(int64_t a, int32x4_t b, int32x4_t c) {
#ifdef POLYMORPHIC
  return vrmlsldavhaxq(a, b, c);
#else
  return vrmlsldavhaxq_s32(a, b, c);
#endif
}

// CHECK-LABEL: @test_vmlaldavaq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]], <8 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlaldavaq_p_s16(int64_t a, int16x8_t b, int16x8_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavaq_p(a, b, c, p);
#else
  return vmlaldavaq_p_s16(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavaq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlaldavaq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavaq_p(a, b, c, p);
#else
  return vmlaldavaq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavaq_p_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 1, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]], <8 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
uint64_t test_vmlaldavaq_p_u16(uint64_t a, uint16x8_t b, uint16x8_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavaq_p(a, b, c, p);
#else
  return vmlaldavaq_p_u16(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavaq_p_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 1, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
uint64_t test_vmlaldavaq_p_u32(uint64_t a, uint32x4_t b, uint32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavaq_p(a, b, c, p);
#else
  return vmlaldavaq_p_u32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavaxq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 0, i32 1, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]], <8 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlaldavaxq_p_s16(int64_t a, int16x8_t b, int16x8_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavaxq_p(a, b, c, p);
#else
  return vmlaldavaxq_p_s16(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavaxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 0, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlaldavaxq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavaxq_p(a, b, c, p);
#else
  return vmlaldavaxq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlsldavaq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 1, i32 0, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]], <8 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlsldavaq_p_s16(int64_t a, int16x8_t b, int16x8_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavaq_p(a, b, c, p);
#else
  return vmlsldavaq_p_s16(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlsldavaq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 1, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlsldavaq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavaq_p(a, b, c, p);
#else
  return vmlsldavaq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlsldaxvaq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 1, i32 1, i32 [[TMP2]], i32 [[TMP1]], <8 x i16> [[B:%.*]], <8 x i16> [[C:%.*]], <8 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlsldaxvaq_p_s16(int64_t a, int16x8_t b, int16x8_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavaxq_p(a, b, c, p);
#else
  return vmlsldavaxq_p_s16(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlsldavaxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 1, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vmlsldavaxq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavaxq_p(a, b, c, p);
#else
  return vmlsldavaxq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhaq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vrmlaldavhaq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlaldavhaq_p(a, b, c, p);
#else
  return vrmlaldavhaq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhaq_p_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 1, i32 0, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
uint64_t test_vrmlaldavhaq_p_u32(uint64_t a, uint32x4_t b, uint32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlaldavhaq_p(a, b, c, p);
#else
  return vrmlaldavhaq_p_u32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhaxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 0, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vrmlaldavhaxq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlaldavhaxq_p(a, b, c, p);
#else
  return vrmlaldavhaxq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhaq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 1, i32 0, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vrmlsldavhaq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlsldavhaq_p(a, b, c, p);
#else
  return vrmlsldavhaq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhaxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = lshr i64 [[A:%.*]], 32
// CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK-NEXT:    [[TMP2:%.*]] = trunc i64 [[A]] to i32
// CHECK-NEXT:    [[TMP3:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP4:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP3]])
// CHECK-NEXT:    [[TMP5:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 1, i32 1, i32 [[TMP2]], i32 [[TMP1]], <4 x i32> [[B:%.*]], <4 x i32> [[C:%.*]], <4 x i1> [[TMP4]])
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP5]], 1
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = shl i64 [[TMP7]], 32
// CHECK-NEXT:    [[TMP9:%.*]] = extractvalue { i32, i32 } [[TMP5]], 0
// CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP8]], [[TMP10]]
// CHECK-NEXT:    ret i64 [[TMP11]]
//
int64_t test_vrmlsldavhaxq_p_s32(int64_t a, int32x4_t b, int32x4_t c, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlsldavhaxq_p(a, b, c, p);
#else
  return vrmlsldavhaxq_p_s32(a, b, c, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 0, i32 0, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlaldavq_s16(int16x8_t a, int16x8_t b) {
#ifdef POLYMORPHIC
  return vmlaldavq(a, b);
#else
  return vmlaldavq_s16(a, b);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlaldavq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vmlaldavq(a, b);
#else
  return vmlaldavq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 1, i32 0, i32 0, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
uint64_t test_vmlaldavq_u16(uint16x8_t a, uint16x8_t b) {
#ifdef POLYMORPHIC
  return vmlaldavq(a, b);
#else
  return vmlaldavq_u16(a, b);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 1, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
uint64_t test_vmlaldavq_u32(uint32x4_t a, uint32x4_t b) {
#ifdef POLYMORPHIC
  return vmlaldavq(a, b);
#else
  return vmlaldavq_u32(a, b);
#endif
}

// CHECK-LABEL: @test_vmlaldavxq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 0, i32 1, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlaldavxq_s16(int16x8_t a, int16x8_t b) {
#ifdef POLYMORPHIC
  return vmlaldavxq(a, b);
#else
  return vmlaldavxq_s16(a, b);
#endif
}

// CHECK-LABEL: @test_vmlaldavxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 0, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlaldavxq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vmlaldavxq(a, b);
#else
  return vmlaldavxq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vmlsldavq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 1, i32 0, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlsldavq_s16(int16x8_t a, int16x8_t b) {
#ifdef POLYMORPHIC
  return vmlsldavq(a, b);
#else
  return vmlsldavq_s16(a, b);
#endif
}

// CHECK-LABEL: @test_vmlsldavq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 1, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlsldavq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vmlsldavq(a, b);
#else
  return vmlsldavq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vmlsldavxvq_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 1, i32 1, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlsldavxvq_s16(int16x8_t a, int16x8_t b) {
#ifdef POLYMORPHIC
  return vmlsldavxq(a, b);
#else
  return vmlsldavxq_s16(a, b);
#endif
}

// CHECK-LABEL: @test_vmlsldavxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.v4i32(i32 0, i32 1, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vmlsldavxq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vmlsldavxq(a, b);
#else
  return vmlsldavxq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vrmlaldavhq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vrmlaldavhq(a, b);
#else
  return vrmlaldavhq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhq_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 1, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
uint64_t test_vrmlaldavhq_u32(uint32x4_t a, uint32x4_t b) {
#ifdef POLYMORPHIC
  return vrmlaldavhq(a, b);
#else
  return vrmlaldavhq_u32(a, b);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 0, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vrmlaldavhxq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vrmlaldavhxq(a, b);
#else
  return vrmlaldavhxq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 1, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vrmlsldavhq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vrmlsldavhq(a, b);
#else
  return vrmlsldavhq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhxq_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.v4i32(i32 0, i32 1, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]])
// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { i32, i32 } [[TMP0]], 1
// CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK-NEXT:    [[TMP3:%.*]] = shl i64 [[TMP2]], 32
// CHECK-NEXT:    [[TMP4:%.*]] = extractvalue { i32, i32 } [[TMP0]], 0
// CHECK-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP3]], [[TMP5]]
// CHECK-NEXT:    ret i64 [[TMP6]]
//
int64_t test_vrmlsldavhxq_s32(int32x4_t a, int32x4_t b) {
#ifdef POLYMORPHIC
  return vrmlsldavhxq(a, b);
#else
  return vrmlsldavhxq_s32(a, b);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 0, i32 0, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]], <8 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlaldavq_p_s16(int16x8_t a, int16x8_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavq_p(a, b, p);
#else
  return vmlaldavq_p_s16(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlaldavq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavq_p(a, b, p);
#else
  return vmlaldavq_p_s32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_p_u16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 1, i32 0, i32 0, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]], <8 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
uint64_t test_vmlaldavq_p_u16(uint16x8_t a, uint16x8_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavq_p(a, b, p);
#else
  return vmlaldavq_p_u16(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavq_p_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 1, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
uint64_t test_vmlaldavq_p_u32(uint32x4_t a, uint32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavq_p(a, b, p);
#else
  return vmlaldavq_p_u32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavxq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 0, i32 1, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]], <8 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlaldavxq_p_s16(int16x8_t a, int16x8_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavxq_p(a, b, p);
#else
  return vmlaldavxq_p_s16(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlaldavxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 0, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlaldavxq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlaldavxq_p(a, b, p);
#else
  return vmlaldavxq_p_s32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlsldavq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 1, i32 0, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]], <8 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlsldavq_p_s16(int16x8_t a, int16x8_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavq_p(a, b, p);
#else
  return vmlsldavq_p_s16(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlsldavq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 1, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlsldavq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavq_p(a, b, p);
#else
  return vmlsldavq_p_s32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlsldaxvq_p_s16(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v8i16.v8i1(i32 0, i32 1, i32 1, i32 0, i32 0, <8 x i16> [[A:%.*]], <8 x i16> [[B:%.*]], <8 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlsldaxvq_p_s16(int16x8_t a, int16x8_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavxq_p(a, b, p);
#else
  return vmlsldavxq_p_s16(a, b, p);
#endif
}

// CHECK-LABEL: @test_vmlsldavxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vmlldava.predicated.v4i32.v4i1(i32 0, i32 1, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vmlsldavxq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vmlsldavxq_p(a, b, p);
#else
  return vmlsldavxq_p_s32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vrmlaldavhq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlaldavhq_p(a, b, p);
#else
  return vrmlaldavhq_p_s32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhq_p_u32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 1, i32 0, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
uint64_t test_vrmlaldavhq_p_u32(uint32x4_t a, uint32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlaldavhq_p(a, b, p);
#else
  return vrmlaldavhq_p_u32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vrmlaldavhxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 0, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vrmlaldavhxq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlaldavhxq_p(a, b, p);
#else
  return vrmlaldavhxq_p_s32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 1, i32 0, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vrmlsldavhq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlsldavhq_p(a, b, p);
#else
  return vrmlsldavhq_p_s32(a, b, p);
#endif
}

// CHECK-LABEL: @test_vrmlsldavhxq_p_s32(
// CHECK-NEXT:  entry:
// CHECK-NEXT:    [[TMP0:%.*]] = zext i16 [[P:%.*]] to i32
// CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 [[TMP0]])
// CHECK-NEXT:    [[TMP2:%.*]] = call { i32, i32 } @llvm.arm.mve.vrmlldavha.predicated.v4i32.v4i1(i32 0, i32 1, i32 1, i32 0, i32 0, <4 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <4 x i1> [[TMP1]])
// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { i32, i32 } [[TMP2]], 1
// CHECK-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP3]] to i64
// CHECK-NEXT:    [[TMP5:%.*]] = shl i64 [[TMP4]], 32
// CHECK-NEXT:    [[TMP6:%.*]] = extractvalue { i32, i32 } [[TMP2]], 0
// CHECK-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP6]] to i64
// CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP5]], [[TMP7]]
// CHECK-NEXT:    ret i64 [[TMP8]]
//
int64_t test_vrmlsldavhxq_p_s32(int32x4_t a, int32x4_t b, mve_pred16_t p) {
#ifdef POLYMORPHIC
  return vrmlsldavhxq_p(a, b, p);
#else
  return vrmlsldavhxq_p_s32(a, b, p);
#endif
}