; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
; RUN: opt --passes=loop-vectorize --mtriple=aarch64 -mattr="+neon,+fullfp16" -S < %s | FileCheck %s
define void @fmin32(ptr noundef readonly captures(none) %input1, ptr noundef readonly captures(none) %input2, ptr noundef writeonly captures(none) %output) {
; CHECK-LABEL: define void @fmin32(
; CHECK-SAME: ptr noundef readonly captures(none) [[INPUT1:%.*]], ptr noundef readonly captures(none) [[INPUT2:%.*]], ptr noundef writeonly captures(none) [[OUTPUT:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[INPUT23:%.*]] = ptrtoint ptr [[INPUT2]] to i64
; CHECK-NEXT:    [[INPUT12:%.*]] = ptrtoint ptr [[INPUT1]] to i64
; CHECK-NEXT:    [[OUTPUT1:%.*]] = ptrtoint ptr [[OUTPUT]] to i64
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[OUTPUT1]], [[INPUT12]]
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = sub i64 [[OUTPUT1]], [[INPUT23]]
; CHECK-NEXT:    [[DIFF_CHECK4:%.*]] = icmp ult i64 [[TMP1]], 32
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[DIFF_CHECK]], [[DIFF_CHECK4]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT1]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw float, ptr [[TMP2]], i64 4
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
; CHECK-NEXT:    [[WIDE_LOAD5:%.*]] = load <4 x float>, ptr [[TMP4]], align 4
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT2]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw float, ptr [[TMP5]], i64 4
; CHECK-NEXT:    [[WIDE_LOAD6:%.*]] = load <4 x float>, ptr [[TMP5]], align 4
; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x float>, ptr [[TMP7]], align 4
; CHECK-NEXT:    [[TMP8:%.*]] = call <4 x float> @llvm.minimumnum.v4f32(<4 x float> [[WIDE_LOAD]], <4 x float> [[WIDE_LOAD6]])
; CHECK-NEXT:    [[TMP9:%.*]] = call <4 x float> @llvm.minimumnum.v4f32(<4 x float> [[WIDE_LOAD5]], <4 x float> [[WIDE_LOAD7]])
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[OUTPUT]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw float, ptr [[TMP10]], i64 4
; CHECK-NEXT:    store <4 x float> [[TMP8]], ptr [[TMP10]], align 4
; CHECK-NEXT:    store <4 x float> [[TMP9]], ptr [[TMP12]], align 4
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 4096
; CHECK-NEXT:    br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    br label %[[FOR_BODY:.*]]
; CHECK:       [[FOR_BODY]]:
; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT1]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN1:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT2]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN2:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT:    [[OUT:%.*]] = tail call float @llvm.minimumnum.f32(float [[IN1]], float [[IN2]])
; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[OUTPUT]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    store float [[OUT]], ptr [[ARRAYIDX4]], align 4
; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 4096
; CHECK-NEXT:    br i1 [[EXITCOND_NOT]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds nuw [4096 x float], ptr %input1, i64 0, i64 %iv
  %in1 = load float, ptr %arrayidx, align 4
  %arrayidx2 = getelementptr inbounds nuw [4096 x float], ptr %input2, i64 0, i64 %iv
  %in2 = load float, ptr %arrayidx2, align 4
  %out = tail call float @llvm.minimumnum.f32(float %in1, float %in2)
  %arrayidx4 = getelementptr inbounds nuw [4096 x float], ptr %output, i64 0, i64 %iv
  store float %out, ptr %arrayidx4, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 4096
  br i1 %exitcond.not, label %exit, label %for.body

exit:
  ret void
}

declare float @llvm.minimumnum.f32(float, float)

define void @fmax32(ptr noundef readonly captures(none) %input1, ptr noundef readonly captures(none) %input2, ptr noundef writeonly captures(none) %output) {
; CHECK-LABEL: define void @fmax32(
; CHECK-SAME: ptr noundef readonly captures(none) [[INPUT1:%.*]], ptr noundef readonly captures(none) [[INPUT2:%.*]], ptr noundef writeonly captures(none) [[OUTPUT:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[INPUT23:%.*]] = ptrtoint ptr [[INPUT2]] to i64
; CHECK-NEXT:    [[INPUT12:%.*]] = ptrtoint ptr [[INPUT1]] to i64
; CHECK-NEXT:    [[OUTPUT1:%.*]] = ptrtoint ptr [[OUTPUT]] to i64
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[OUTPUT1]], [[INPUT12]]
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = sub i64 [[OUTPUT1]], [[INPUT23]]
; CHECK-NEXT:    [[DIFF_CHECK4:%.*]] = icmp ult i64 [[TMP1]], 32
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[DIFF_CHECK]], [[DIFF_CHECK4]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT1]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw float, ptr [[TMP2]], i64 4
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
; CHECK-NEXT:    [[WIDE_LOAD5:%.*]] = load <4 x float>, ptr [[TMP4]], align 4
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT2]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw float, ptr [[TMP5]], i64 4
; CHECK-NEXT:    [[WIDE_LOAD6:%.*]] = load <4 x float>, ptr [[TMP5]], align 4
; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x float>, ptr [[TMP7]], align 4
; CHECK-NEXT:    [[TMP8:%.*]] = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> [[WIDE_LOAD]], <4 x float> [[WIDE_LOAD6]])
; CHECK-NEXT:    [[TMP9:%.*]] = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> [[WIDE_LOAD5]], <4 x float> [[WIDE_LOAD7]])
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[OUTPUT]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw float, ptr [[TMP10]], i64 4
; CHECK-NEXT:    store <4 x float> [[TMP8]], ptr [[TMP10]], align 4
; CHECK-NEXT:    store <4 x float> [[TMP9]], ptr [[TMP12]], align 4
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 4096
; CHECK-NEXT:    br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    br label %[[FOR_BODY:.*]]
; CHECK:       [[FOR_BODY]]:
; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT1]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN1:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[INPUT2]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN2:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT:    [[OUT:%.*]] = tail call float @llvm.maximumnum.f32(float [[IN1]], float [[IN2]])
; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw [4096 x float], ptr [[OUTPUT]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    store float [[OUT]], ptr [[ARRAYIDX4]], align 4
; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 4096
; CHECK-NEXT:    br i1 [[EXITCOND_NOT]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds nuw [4096 x float], ptr %input1, i64 0, i64 %iv
  %in1 = load float, ptr %arrayidx, align 4
  %arrayidx2 = getelementptr inbounds nuw [4096 x float], ptr %input2, i64 0, i64 %iv
  %in2 = load float, ptr %arrayidx2, align 4
  %out = tail call float @llvm.maximumnum.f32(float %in1, float %in2)
  %arrayidx4 = getelementptr inbounds nuw [4096 x float], ptr %output, i64 0, i64 %iv
  store float %out, ptr %arrayidx4, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 4096
  br i1 %exitcond.not, label %exit, label %for.body

exit:
  ret void
}

declare float @llvm.maximumnum.f32(float, float)

define void @fmin64(ptr noundef readonly captures(none) %input1, ptr noundef readonly captures(none) %input2, ptr noundef writeonly captures(none) %output) {
; CHECK-LABEL: define void @fmin64(
; CHECK-SAME: ptr noundef readonly captures(none) [[INPUT1:%.*]], ptr noundef readonly captures(none) [[INPUT2:%.*]], ptr noundef writeonly captures(none) [[OUTPUT:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[INPUT23:%.*]] = ptrtoint ptr [[INPUT2]] to i64
; CHECK-NEXT:    [[INPUT12:%.*]] = ptrtoint ptr [[INPUT1]] to i64
; CHECK-NEXT:    [[OUTPUT1:%.*]] = ptrtoint ptr [[OUTPUT]] to i64
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[OUTPUT1]], [[INPUT12]]
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = sub i64 [[OUTPUT1]], [[INPUT23]]
; CHECK-NEXT:    [[DIFF_CHECK4:%.*]] = icmp ult i64 [[TMP1]], 32
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[DIFF_CHECK]], [[DIFF_CHECK4]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT1]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw double, ptr [[TMP2]], i64 2
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP2]], align 8
; CHECK-NEXT:    [[WIDE_LOAD5:%.*]] = load <2 x double>, ptr [[TMP4]], align 8
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT2]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw double, ptr [[TMP5]], i64 2
; CHECK-NEXT:    [[WIDE_LOAD6:%.*]] = load <2 x double>, ptr [[TMP5]], align 8
; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <2 x double>, ptr [[TMP7]], align 8
; CHECK-NEXT:    [[TMP8:%.*]] = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> [[WIDE_LOAD]], <2 x double> [[WIDE_LOAD6]])
; CHECK-NEXT:    [[TMP9:%.*]] = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> [[WIDE_LOAD5]], <2 x double> [[WIDE_LOAD7]])
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[OUTPUT]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw double, ptr [[TMP10]], i64 2
; CHECK-NEXT:    store <2 x double> [[TMP8]], ptr [[TMP10]], align 8
; CHECK-NEXT:    store <2 x double> [[TMP9]], ptr [[TMP12]], align 8
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 4096
; CHECK-NEXT:    br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    br label %[[FOR_BODY:.*]]
; CHECK:       [[FOR_BODY]]:
; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT1]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN1:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT2]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN2:%.*]] = load double, ptr [[ARRAYIDX2]], align 8
; CHECK-NEXT:    [[OUT:%.*]] = tail call double @llvm.minimumnum.f64(double [[IN1]], double [[IN2]])
; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[OUTPUT]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    store double [[OUT]], ptr [[ARRAYIDX4]], align 8
; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 4096
; CHECK-NEXT:    br i1 [[EXITCOND_NOT]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds nuw [4096 x double], ptr %input1, i64 0, i64 %iv
  %in1 = load double, ptr %arrayidx, align 8
  %arrayidx2 = getelementptr inbounds nuw [4096 x double], ptr %input2, i64 0, i64 %iv
  %in2 = load double, ptr %arrayidx2, align 8
  %out = tail call double @llvm.minimumnum.f64(double %in1, double %in2)
  %arrayidx4 = getelementptr inbounds nuw [4096 x double], ptr %output, i64 0, i64 %iv
  store double %out, ptr %arrayidx4, align 8
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 4096
  br i1 %exitcond.not, label %exit, label %for.body

exit:
  ret void
}

declare double @llvm.minimumnum.f64(double, double)

define void @fmax64(ptr noundef readonly captures(none) %input1, ptr noundef readonly captures(none) %input2, ptr noundef writeonly captures(none) %output) {
; CHECK-LABEL: define void @fmax64(
; CHECK-SAME: ptr noundef readonly captures(none) [[INPUT1:%.*]], ptr noundef readonly captures(none) [[INPUT2:%.*]], ptr noundef writeonly captures(none) [[OUTPUT:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[INPUT23:%.*]] = ptrtoint ptr [[INPUT2]] to i64
; CHECK-NEXT:    [[INPUT12:%.*]] = ptrtoint ptr [[INPUT1]] to i64
; CHECK-NEXT:    [[OUTPUT1:%.*]] = ptrtoint ptr [[OUTPUT]] to i64
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[OUTPUT1]], [[INPUT12]]
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = sub i64 [[OUTPUT1]], [[INPUT23]]
; CHECK-NEXT:    [[DIFF_CHECK4:%.*]] = icmp ult i64 [[TMP1]], 32
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[DIFF_CHECK]], [[DIFF_CHECK4]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT1]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw double, ptr [[TMP2]], i64 2
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP2]], align 8
; CHECK-NEXT:    [[WIDE_LOAD5:%.*]] = load <2 x double>, ptr [[TMP4]], align 8
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT2]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw double, ptr [[TMP5]], i64 2
; CHECK-NEXT:    [[WIDE_LOAD6:%.*]] = load <2 x double>, ptr [[TMP5]], align 8
; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <2 x double>, ptr [[TMP7]], align 8
; CHECK-NEXT:    [[TMP8:%.*]] = call <2 x double> @llvm.maximumnum.v2f64(<2 x double> [[WIDE_LOAD]], <2 x double> [[WIDE_LOAD6]])
; CHECK-NEXT:    [[TMP9:%.*]] = call <2 x double> @llvm.maximumnum.v2f64(<2 x double> [[WIDE_LOAD5]], <2 x double> [[WIDE_LOAD7]])
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[OUTPUT]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw double, ptr [[TMP10]], i64 2
; CHECK-NEXT:    store <2 x double> [[TMP8]], ptr [[TMP10]], align 8
; CHECK-NEXT:    store <2 x double> [[TMP9]], ptr [[TMP12]], align 8
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 4096
; CHECK-NEXT:    br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    br label %[[FOR_BODY:.*]]
; CHECK:       [[FOR_BODY]]:
; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT1]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN1:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[INPUT2]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN2:%.*]] = load double, ptr [[ARRAYIDX2]], align 8
; CHECK-NEXT:    [[OUT:%.*]] = tail call double @llvm.maximumnum.f64(double [[IN1]], double [[IN2]])
; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw [4096 x double], ptr [[OUTPUT]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    store double [[OUT]], ptr [[ARRAYIDX4]], align 8
; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 4096
; CHECK-NEXT:    br i1 [[EXITCOND_NOT]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds nuw [4096 x double], ptr %input1, i64 0, i64 %iv
  %in1 = load double, ptr %arrayidx, align 8
  %arrayidx2 = getelementptr inbounds nuw [4096 x double], ptr %input2, i64 0, i64 %iv
  %in2 = load double, ptr %arrayidx2, align 8
  %out = tail call double @llvm.maximumnum.f64(double %in1, double %in2)
  %arrayidx4 = getelementptr inbounds nuw [4096 x double], ptr %output, i64 0, i64 %iv
  store double %out, ptr %arrayidx4, align 8
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 4096
  br i1 %exitcond.not, label %exit, label %for.body

exit:
  ret void
}

declare double @llvm.maximumnum.f64(double, double)

define void @fmin16(ptr noundef readonly captures(none) %input1, ptr noundef readonly captures(none) %input2, ptr noundef writeonly captures(none) %output) {
; CHECK-LABEL: define void @fmin16(
; CHECK-SAME: ptr noundef readonly captures(none) [[INPUT1:%.*]], ptr noundef readonly captures(none) [[INPUT2:%.*]], ptr noundef writeonly captures(none) [[OUTPUT:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[INPUT23:%.*]] = ptrtoint ptr [[INPUT2]] to i64
; CHECK-NEXT:    [[INPUT12:%.*]] = ptrtoint ptr [[INPUT1]] to i64
; CHECK-NEXT:    [[OUTPUT1:%.*]] = ptrtoint ptr [[OUTPUT]] to i64
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[OUTPUT1]], [[INPUT12]]
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = sub i64 [[OUTPUT1]], [[INPUT23]]
; CHECK-NEXT:    [[DIFF_CHECK4:%.*]] = icmp ult i64 [[TMP1]], 32
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[DIFF_CHECK]], [[DIFF_CHECK4]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT1]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw half, ptr [[TMP2]], i64 8
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <8 x half>, ptr [[TMP2]], align 2
; CHECK-NEXT:    [[WIDE_LOAD5:%.*]] = load <8 x half>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT2]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw half, ptr [[TMP4]], i64 8
; CHECK-NEXT:    [[WIDE_LOAD6:%.*]] = load <8 x half>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <8 x half>, ptr [[TMP10]], align 2
; CHECK-NEXT:    [[TMP11:%.*]] = call <8 x half> @llvm.minimumnum.v8f16(<8 x half> [[WIDE_LOAD]], <8 x half> [[WIDE_LOAD6]])
; CHECK-NEXT:    [[TMP13:%.*]] = call <8 x half> @llvm.minimumnum.v8f16(<8 x half> [[WIDE_LOAD5]], <8 x half> [[WIDE_LOAD7]])
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[OUTPUT]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw half, ptr [[TMP7]], i64 8
; CHECK-NEXT:    store <8 x half> [[TMP11]], ptr [[TMP7]], align 2
; CHECK-NEXT:    store <8 x half> [[TMP13]], ptr [[TMP12]], align 2
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 4096
; CHECK-NEXT:    br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    br label %[[FOR_BODY:.*]]
; CHECK:       [[FOR_BODY]]:
; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT1]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN1:%.*]] = load half, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT2]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN2:%.*]] = load half, ptr [[ARRAYIDX2]], align 2
; CHECK-NEXT:    [[OUT:%.*]] = tail call half @llvm.minimumnum.f16(half [[IN1]], half [[IN2]])
; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[OUTPUT]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    store half [[OUT]], ptr [[ARRAYIDX4]], align 2
; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 4096
; CHECK-NEXT:    br i1 [[EXITCOND_NOT]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds nuw [4096 x half], ptr %input1, i64 0, i64 %iv
  %in1 = load half, ptr %arrayidx, align 2
  %arrayidx2 = getelementptr inbounds nuw [4096 x half], ptr %input2, i64 0, i64 %iv
  %in2 = load half, ptr %arrayidx2, align 2
  %out = tail call half @llvm.minimumnum.f16(half %in1, half %in2)
  %arrayidx4 = getelementptr inbounds nuw [4096 x half], ptr %output, i64 0, i64 %iv
  store half %out, ptr %arrayidx4, align 2
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 4096
  br i1 %exitcond.not, label %exit, label %for.body

exit:
  ret void
}

declare half @llvm.minimumnum.f16(half, half)

define void @fmax16(ptr noundef readonly captures(none) %input1, ptr noundef readonly captures(none) %input2, ptr noundef writeonly captures(none) %output) {
; CHECK-LABEL: define void @fmax16(
; CHECK-SAME: ptr noundef readonly captures(none) [[INPUT1:%.*]], ptr noundef readonly captures(none) [[INPUT2:%.*]], ptr noundef writeonly captures(none) [[OUTPUT:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[INPUT23:%.*]] = ptrtoint ptr [[INPUT2]] to i64
; CHECK-NEXT:    [[INPUT12:%.*]] = ptrtoint ptr [[INPUT1]] to i64
; CHECK-NEXT:    [[OUTPUT1:%.*]] = ptrtoint ptr [[OUTPUT]] to i64
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[OUTPUT1]], [[INPUT12]]
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = sub i64 [[OUTPUT1]], [[INPUT23]]
; CHECK-NEXT:    [[DIFF_CHECK4:%.*]] = icmp ult i64 [[TMP1]], 32
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[DIFF_CHECK]], [[DIFF_CHECK4]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT1]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw half, ptr [[TMP2]], i64 8
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <8 x half>, ptr [[TMP2]], align 2
; CHECK-NEXT:    [[WIDE_LOAD5:%.*]] = load <8 x half>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT2]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw half, ptr [[TMP4]], i64 8
; CHECK-NEXT:    [[WIDE_LOAD6:%.*]] = load <8 x half>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[WIDE_LOAD7:%.*]] = load <8 x half>, ptr [[TMP10]], align 2
; CHECK-NEXT:    [[TMP11:%.*]] = call <8 x half> @llvm.maximumnum.v8f16(<8 x half> [[WIDE_LOAD]], <8 x half> [[WIDE_LOAD6]])
; CHECK-NEXT:    [[TMP13:%.*]] = call <8 x half> @llvm.maximumnum.v8f16(<8 x half> [[WIDE_LOAD5]], <8 x half> [[WIDE_LOAD7]])
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[OUTPUT]], i64 0, i64 [[INDEX]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw half, ptr [[TMP7]], i64 8
; CHECK-NEXT:    store <8 x half> [[TMP11]], ptr [[TMP7]], align 2
; CHECK-NEXT:    store <8 x half> [[TMP13]], ptr [[TMP12]], align 2
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 4096
; CHECK-NEXT:    br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    br label %[[FOR_BODY:.*]]
; CHECK:       [[FOR_BODY]]:
; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT1]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN1:%.*]] = load half, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[INPUT2]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    [[IN2:%.*]] = load half, ptr [[ARRAYIDX2]], align 2
; CHECK-NEXT:    [[OUT:%.*]] = tail call half @llvm.maximumnum.f16(half [[IN1]], half [[IN2]])
; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw [4096 x half], ptr [[OUTPUT]], i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT:    store half [[OUT]], ptr [[ARRAYIDX4]], align 2
; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 4096
; CHECK-NEXT:    br i1 [[EXITCOND_NOT]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds nuw [4096 x half], ptr %input1, i64 0, i64 %iv
  %in1 = load half, ptr %arrayidx, align 2
  %arrayidx2 = getelementptr inbounds nuw [4096 x half], ptr %input2, i64 0, i64 %iv
  %in2 = load half, ptr %arrayidx2, align 2
  %out = tail call half @llvm.maximumnum.f16(half %in1, half %in2)
  %arrayidx4 = getelementptr inbounds nuw [4096 x half], ptr %output, i64 0, i64 %iv
  store half %out, ptr %arrayidx4, align 2
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 4096
  br i1 %exitcond.not, label %exit, label %for.body

exit:
  ret void
}

declare half @llvm.maximumnum.f16(half, half)
;.
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META1]]}
; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META1]]}
; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META1]], [[META2]]}
; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META1]]}
; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META1]], [[META2]]}
; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META1]]}
; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META1]], [[META2]]}
; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META1]]}
; CHECK: [[LOOP12]] = distinct !{[[LOOP12]], [[META1]], [[META2]]}
; CHECK: [[LOOP13]] = distinct !{[[LOOP13]], [[META1]]}
;.