// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
// RUN: %clang_cc1        -triple aarch64-none-linux-gnu -target-feature +neon -target-feature +bf16 -target-feature +fp8 -target-feature +fp8fma -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,sroa | FileCheck %s
// RUN: %clang_cc1 -x c++ -triple aarch64-none-linux-gnu -target-feature +neon -target-feature +bf16 -target-feature +fp8 -target-feature +fp8fma -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,sroa | FileCheck %s -check-prefix CHECK-CXX

// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +neon -target-feature +bf16 -target-feature +fp8 -target-feature +fp8fma -disable-O0-optnone -Werror -Wall -S -o /dev/null %s

// REQUIRES: aarch64-registered-target

#include <arm_neon.h>

// CHECK-LABEL: define dso_local <8 x half> @test_vmlalb(
// CHECK-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLAL_I:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-NEXT:    [[VMLAL1_I:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalb.v8f16(<8 x half> [[VMLAL_I]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-NEXT:    ret <8 x half> [[VMLAL1_I]]
//
// CHECK-CXX-LABEL: define dso_local noundef <8 x half> @_Z11test_vmlalb13__Float16x8_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-CXX-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLAL_I:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-CXX-NEXT:    [[VMLAL1_I:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalb.v8f16(<8 x half> [[VMLAL_I]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-CXX-NEXT:    ret <8 x half> [[VMLAL1_I]]
//
float16x8_t test_vmlalb(float16x8_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlalbq_f16_mf8_fpm(vd, vn, vm, fpm);
}

// CHECK-LABEL: define dso_local <8 x half> @test_vmlalt(
// CHECK-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLAL_I:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-NEXT:    [[VMLAL1_I:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalt.v8f16(<8 x half> [[VMLAL_I]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-NEXT:    ret <8 x half> [[VMLAL1_I]]
//
// CHECK-CXX-LABEL: define dso_local noundef <8 x half> @_Z11test_vmlalt13__Float16x8_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-CXX-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLAL_I:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-CXX-NEXT:    [[VMLAL1_I:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalt.v8f16(<8 x half> [[VMLAL_I]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-CXX-NEXT:    ret <8 x half> [[VMLAL1_I]]
//
float16x8_t test_vmlalt(float16x8_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlaltq_f16_mf8_fpm(vd, vn, vm, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlallbb(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbb.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-NEXT:    ret <4 x float> [[VMLALL_I]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z13test_vmlallbb13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbb.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_I]]
//
float32x4_t test_vmlallbb(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlallbbq_f32_mf8_fpm(vd, vn, vm, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlallbt(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbt.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-NEXT:    ret <4 x float> [[VMLALL_I]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z13test_vmlallbt13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbt.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_I]]
//
float32x4_t test_vmlallbt(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlallbtq_f32_mf8_fpm(vd, vn, vm, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlalltb(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltb.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-NEXT:    ret <4 x float> [[VMLALL_I]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z13test_vmlalltb13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltb.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_I]]
//
float32x4_t test_vmlalltb(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlalltbq_f32_mf8_fpm(vd, vn, vm, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlalltt(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltt.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-NEXT:    ret <4 x float> [[VMLALL_I]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z13test_vmlalltt13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltt.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]])
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_I]]
//
float32x4_t test_vmlalltt(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlallttq_f32_mf8_fpm(vd, vn, vm, fpm);
}

// CHECK-LABEL: define dso_local <8 x half> @test_vmlalb_lane(
// CHECK-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-NEXT:    [[TMP2:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalb.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[TMP2]], i32 0)
// CHECK-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
// CHECK-CXX-LABEL: define dso_local noundef <8 x half> @_Z16test_vmlalb_lane13__Float16x8_t14__Mfloat8x16_t13__Mfloat8x8_tm(
// CHECK-CXX-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-CXX-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-CXX-NEXT:    [[TMP2:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-CXX-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalb.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[TMP2]], i32 0)
// CHECK-CXX-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
float16x8_t test_vmlalb_lane(float16x8_t vd, mfloat8x16_t vn, mfloat8x8_t vm, fpm_t fpm) {
  return vmlalbq_lane_f16_mf8_fpm(vd, vn, vm, 0, fpm);
}

// CHECK-LABEL: define dso_local <8 x half> @test_vmlalb_laneq(
// CHECK-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalb.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 0)
// CHECK-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
// CHECK-CXX-LABEL: define dso_local noundef <8 x half> @_Z17test_vmlalb_laneq13__Float16x8_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-CXX-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-CXX-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalb.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 0)
// CHECK-CXX-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
float16x8_t test_vmlalb_laneq(float16x8_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlalbq_laneq_f16_mf8_fpm(vd, vn, vm, 0, fpm);
}

// CHECK-LABEL: define dso_local <8 x half> @test_vmlalt_lane(
// CHECK-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-NEXT:    [[TMP2:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalt.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[TMP2]], i32 7)
// CHECK-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
// CHECK-CXX-LABEL: define dso_local noundef <8 x half> @_Z16test_vmlalt_lane13__Float16x8_t14__Mfloat8x16_t13__Mfloat8x8_tm(
// CHECK-CXX-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-CXX-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-CXX-NEXT:    [[TMP2:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-CXX-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalt.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[TMP2]], i32 7)
// CHECK-CXX-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
float16x8_t test_vmlalt_lane(float16x8_t vd, mfloat8x16_t vn, mfloat8x8_t vm, fpm_t fpm) {
  return vmlaltq_lane_f16_mf8_fpm(vd, vn, vm, 7, fpm);
}

// CHECK-LABEL: define dso_local <8 x half> @test_vmlalt_laneq(
// CHECK-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalt.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 15)
// CHECK-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
// CHECK-CXX-LABEL: define dso_local noundef <8 x half> @_Z17test_vmlalt_laneq13__Float16x8_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <8 x half> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = bitcast <8 x half> [[VD]] to <8 x i16>
// CHECK-CXX-NEXT:    [[TMP1:%.*]] = bitcast <8 x i16> [[TMP0]] to <16 x i8>
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLAL_LANE:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
// CHECK-CXX-NEXT:    [[VMLAL_LANE1:%.*]] = call <8 x half> @llvm.aarch64.neon.fp8.fmlalt.lane.v8f16(<8 x half> [[VMLAL_LANE]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 15)
// CHECK-CXX-NEXT:    ret <8 x half> [[VMLAL_LANE1]]
//
float16x8_t test_vmlalt_laneq(float16x8_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlaltq_laneq_f16_mf8_fpm(vd, vn, vm, 15, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlallbb_lane(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 0)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z18test_vmlallbb_lane13__Float32x4_t14__Mfloat8x16_t13__Mfloat8x8_tm(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 0)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlallbb_lane(float32x4_t vd, mfloat8x16_t vn, mfloat8x8_t vm, fpm_t fpm) {
  return vmlallbbq_lane_f32_mf8_fpm(vd, vn, vm, 0, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlallbb_laneq(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 0)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z19test_vmlallbb_laneq13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 0)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlallbb_laneq(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlallbbq_laneq_f32_mf8_fpm(vd, vn, vm, 0, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlallbt_lane(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 3)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z18test_vmlallbt_lane13__Float32x4_t14__Mfloat8x16_t13__Mfloat8x8_tm(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 3)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlallbt_lane(float32x4_t vd, mfloat8x16_t vn, mfloat8x8_t vm, fpm_t fpm) {
  return vmlallbtq_lane_f32_mf8_fpm(vd, vn, vm, 3, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlallbt_laneq(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 3)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z19test_vmlallbt_laneq13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlallbt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 3)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlallbt_laneq(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlallbtq_laneq_f32_mf8_fpm(vd, vn, vm, 3, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlalltb_lane(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 7)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z18test_vmlalltb_lane13__Float32x4_t14__Mfloat8x16_t13__Mfloat8x8_tm(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 7)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlalltb_lane(float32x4_t vd, mfloat8x16_t vn, mfloat8x8_t vm, fpm_t fpm) {
  return vmlalltbq_lane_f32_mf8_fpm(vd, vn, vm, 7, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlalltb_laneq(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 7)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z19test_vmlalltb_laneq13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltb.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 7)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlalltb_laneq(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlalltbq_laneq_f32_mf8_fpm(vd, vn, vm, 7, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlalltt_lane(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 7)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z18test_vmlalltt_lane13__Float32x4_t14__Mfloat8x16_t13__Mfloat8x8_tm(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <8 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    [[TMP0:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v8i8(<16 x i8> poison, <8 x i8> [[VM]], i64 0)
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[TMP0]], i32 7)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlalltt_lane(float32x4_t vd, mfloat8x16_t vn, mfloat8x8_t vm, fpm_t fpm) {
  return vmlallttq_lane_f32_mf8_fpm(vd, vn, vm, 7, fpm);
}

// CHECK-LABEL: define dso_local <4 x float> @test_vmlalltt_laneq(
// CHECK-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 15)
// CHECK-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z19test_vmlalltt_laneq13__Float32x4_t14__Mfloat8x16_tS0_m(
// CHECK-CXX-SAME: <4 x float> noundef [[VD:%.*]], <16 x i8> [[VN:%.*]], <16 x i8> [[VM:%.*]], i64 noundef [[FPM:%.*]]) #[[ATTR0]] {
// CHECK-CXX-NEXT:  [[ENTRY:.*:]]
// CHECK-CXX-NEXT:    call void @llvm.aarch64.set.fpmr(i64 [[FPM]])
// CHECK-CXX-NEXT:    [[VMLALL_LANE:%.*]] = call <4 x float> @llvm.aarch64.neon.fp8.fmlalltt.lane.v4f32(<4 x float> [[VD]], <16 x i8> [[VN]], <16 x i8> [[VM]], i32 15)
// CHECK-CXX-NEXT:    ret <4 x float> [[VMLALL_LANE]]
//
float32x4_t test_vmlalltt_laneq(float32x4_t vd, mfloat8x16_t vn, mfloat8x16_t vm, fpm_t fpm) {
  return vmlallttq_laneq_f32_mf8_fpm(vd, vn, vm, 15, fpm);
}