; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
; RUN: opt < %s -passes=loop-vectorize -mtriple aarch64-unknown-linux-gnu -mattr=+sve -prefer-predicate-over-epilogue=scalar-epilogue \
; RUN: -force-ordered-reductions=false -hints-allow-reordering=false -S | FileCheck %s --check-prefix=CHECK-NOT-VECTORIZED
; RUN: opt < %s -passes=loop-vectorize -mtriple aarch64-unknown-linux-gnu -mattr=+sve -prefer-predicate-over-epilogue=scalar-epilogue \
; RUN: -force-ordered-reductions=false -hints-allow-reordering=true -S | FileCheck %s --check-prefix=CHECK-UNORDERED
; RUN: opt < %s -passes=loop-vectorize -mtriple aarch64-unknown-linux-gnu -mattr=+sve -prefer-predicate-over-epilogue=scalar-epilogue \
; RUN: -force-ordered-reductions=true -hints-allow-reordering=false -S | FileCheck %s --check-prefix=CHECK-ORDERED
; RUN: opt < %s -passes=loop-vectorize -mtriple aarch64-unknown-linux-gnu -mattr=+sve -prefer-predicate-over-epilogue=scalar-epilogue \
; RUN: -force-ordered-reductions=true -hints-allow-reordering=true -S | FileCheck %s --check-prefix=CHECK-UNORDERED
; RUN: opt < %s -passes=loop-vectorize -mtriple aarch64-unknown-linux-gnu -mattr=+sve -prefer-predicate-over-epilogue=scalar-epilogue \
; RUN: -hints-allow-reordering=false -S | FileCheck %s --check-prefix=CHECK-ORDERED
; RUN: opt < %s -passes=loop-vectorize -mtriple aarch64-unknown-linux-gnu -mattr=+sve -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
; RUN: -hints-allow-reordering=false -S | FileCheck %s --check-prefix=CHECK-ORDERED-TF
define float @fadd_strict(ptr noalias nocapture readonly %a, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define float @fadd_strict
; CHECK-NOT-VECTORIZED-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[SUM_07:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ADD]] = fadd float [[TMP0]], [[SUM_07]]
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: ret float [[ADD_LCSSA]]
;
; CHECK-UNORDERED-LABEL: define float @fadd_strict
; CHECK-UNORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 8
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 8
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 8
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 8 x float> [ insertelement (<vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), float 0.000000e+00, i32 0), [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-UNORDERED-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP7]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP8]], align 4
; CHECK-UNORDERED-NEXT: [[TMP9]] = fadd <vscale x 8 x float> [[WIDE_LOAD]], [[VEC_PHI]]
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-UNORDERED-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[TMP11:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float -0.000000e+00, <vscale x 8 x float> [[TMP9]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP11]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP12:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-UNORDERED-NEXT: [[ADD]] = fadd float [[TMP12]], [[SUM_07]]
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ], [ [[TMP11]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: ret float [[ADD_LCSSA]]
;
; CHECK-ORDERED-LABEL: define float @fadd_strict
; CHECK-ORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 8
; CHECK-ORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-ORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED: vector.ph:
; CHECK-ORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 8
; CHECK-ORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-ORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-ORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 8
; CHECK-ORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED: vector.body:
; CHECK-ORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP7]], i32 0
; CHECK-ORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP8]], align 4
; CHECK-ORDERED-NEXT: [[TMP9]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[WIDE_LOAD]])
; CHECK-ORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-ORDERED-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK-ORDERED: middle.block:
; CHECK-ORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED: scalar.ph:
; CHECK-ORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP9]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP11:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-NEXT: [[ADD]] = fadd float [[TMP11]], [[SUM_07]]
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ], [ [[TMP9]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: ret float [[ADD_LCSSA]]
;
; CHECK-ORDERED-TF-LABEL: define float @fadd_strict
; CHECK-ORDERED-TF-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED-TF: vector.ph:
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP2:%.*]] = sub i64 [[TMP1]], 1
; CHECK-ORDERED-TF-NEXT: [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP2]]
; CHECK-ORDERED-TF-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP1]]
; CHECK-ORDERED-TF-NEXT: [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
; CHECK-ORDERED-TF-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP7:%.*]] = sub i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP8:%.*]] = icmp ugt i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP9:%.*]] = select i1 [[TMP8]], i64 [[TMP7]], i64 0
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 0, i64 [[N]])
; CHECK-ORDERED-TF-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED-TF: vector.body:
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP10]]
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP11]], i32 0
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP12]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP13:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> [[WIDE_MASKED_LOAD]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP14]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[TMP13]])
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = xor <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 8 x i1> insertelement (<vscale x 8 x i1> poison, i1 true, i64 0), <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = extractelement <vscale x 8 x i1> [[TMP15]], i32 0
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK-ORDERED-TF: middle.block:
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED-TF: scalar.ph:
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP14]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP17:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-TF-NEXT: [[ADD]] = fadd float [[TMP17]], [[SUM_07]]
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ], [ [[TMP14]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: ret float [[ADD_LCSSA]]
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%sum.07 = phi float [ 0.000000e+00, %entry ], [ %add, %for.body ]
%arrayidx = getelementptr inbounds float, ptr %a, i64 %iv
%0 = load float, ptr %arrayidx, align 4
%add = fadd float %0, %sum.07
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !0
for.end:
ret float %add
}
define float @fadd_strict_unroll(ptr noalias nocapture readonly %a, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define float @fadd_strict_unroll
; CHECK-NOT-VECTORIZED-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[SUM_07:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ADD]] = fadd float [[TMP0]], [[SUM_07]]
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: ret float [[ADD_LCSSA]]
;
; CHECK-UNORDERED-LABEL: define float @fadd_strict_unroll
; CHECK-UNORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 32
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 32
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 8 x float> [ insertelement (<vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), float 0.000000e+00, i32 0), [[VECTOR_PH]] ], [ [[TMP36:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP37:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI2:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP38:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI3:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP39:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-UNORDERED-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i32 0
; CHECK-UNORDERED-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP28:%.*]] = mul i64 [[TMP27]], 8
; CHECK-UNORDERED-NEXT: [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP28]]
; CHECK-UNORDERED-NEXT: [[TMP30:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP31:%.*]] = mul i64 [[TMP30]], 16
; CHECK-UNORDERED-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP31]]
; CHECK-UNORDERED-NEXT: [[TMP33:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP34:%.*]] = mul i64 [[TMP33]], 24
; CHECK-UNORDERED-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP34]]
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP26]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD4:%.*]] = load <vscale x 8 x float>, ptr [[TMP29]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD5:%.*]] = load <vscale x 8 x float>, ptr [[TMP32]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 8 x float>, ptr [[TMP35]], align 4
; CHECK-UNORDERED-NEXT: [[TMP36]] = fadd <vscale x 8 x float> [[WIDE_LOAD]], [[VEC_PHI]]
; CHECK-UNORDERED-NEXT: [[TMP37]] = fadd <vscale x 8 x float> [[WIDE_LOAD4]], [[VEC_PHI1]]
; CHECK-UNORDERED-NEXT: [[TMP38]] = fadd <vscale x 8 x float> [[WIDE_LOAD5]], [[VEC_PHI2]]
; CHECK-UNORDERED-NEXT: [[TMP39]] = fadd <vscale x 8 x float> [[WIDE_LOAD6]], [[VEC_PHI3]]
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-UNORDERED-NEXT: [[TMP40:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP40]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[BIN_RDX:%.*]] = fadd <vscale x 8 x float> [[TMP37]], [[TMP36]]
; CHECK-UNORDERED-NEXT: [[BIN_RDX7:%.*]] = fadd <vscale x 8 x float> [[TMP38]], [[BIN_RDX]]
; CHECK-UNORDERED-NEXT: [[BIN_RDX8:%.*]] = fadd <vscale x 8 x float> [[TMP39]], [[BIN_RDX7]]
; CHECK-UNORDERED-NEXT: [[TMP41:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float -0.000000e+00, <vscale x 8 x float> [[BIN_RDX8]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP41]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP42:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-UNORDERED-NEXT: [[ADD]] = fadd float [[TMP42]], [[SUM_07]]
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ], [ [[TMP41]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: ret float [[ADD_LCSSA]]
;
; CHECK-ORDERED-LABEL: define float @fadd_strict_unroll
; CHECK-ORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-ORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-ORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED: vector.ph:
; CHECK-ORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 32
; CHECK-ORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-ORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-ORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 32
; CHECK-ORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED: vector.body:
; CHECK-ORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP39:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i32 0
; CHECK-ORDERED-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP28:%.*]] = mul i64 [[TMP27]], 8
; CHECK-ORDERED-NEXT: [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP28]]
; CHECK-ORDERED-NEXT: [[TMP30:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP31:%.*]] = mul i64 [[TMP30]], 16
; CHECK-ORDERED-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP31]]
; CHECK-ORDERED-NEXT: [[TMP33:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP34:%.*]] = mul i64 [[TMP33]], 24
; CHECK-ORDERED-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP34]]
; CHECK-ORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP26]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 8 x float>, ptr [[TMP29]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 8 x float>, ptr [[TMP32]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD3:%.*]] = load <vscale x 8 x float>, ptr [[TMP35]], align 4
; CHECK-ORDERED-NEXT: [[TMP36:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[WIDE_LOAD]])
; CHECK-ORDERED-NEXT: [[TMP37:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP36]], <vscale x 8 x float> [[WIDE_LOAD1]])
; CHECK-ORDERED-NEXT: [[TMP38:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP37]], <vscale x 8 x float> [[WIDE_LOAD2]])
; CHECK-ORDERED-NEXT: [[TMP39]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP38]], <vscale x 8 x float> [[WIDE_LOAD3]])
; CHECK-ORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-ORDERED-NEXT: [[TMP40:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[TMP40]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK-ORDERED: middle.block:
; CHECK-ORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED: scalar.ph:
; CHECK-ORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP39]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP41:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-NEXT: [[ADD]] = fadd float [[TMP41]], [[SUM_07]]
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ], [ [[TMP39]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: ret float [[ADD_LCSSA]]
;
; CHECK-ORDERED-TF-LABEL: define float @fadd_strict_unroll
; CHECK-ORDERED-TF-SAME: (ptr noalias nocapture readonly [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED-TF: vector.ph:
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP2:%.*]] = sub i64 [[TMP1]], 1
; CHECK-ORDERED-TF-NEXT: [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP2]]
; CHECK-ORDERED-TF-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP1]]
; CHECK-ORDERED-TF-NEXT: [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
; CHECK-ORDERED-TF-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP7:%.*]] = sub i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP8:%.*]] = icmp ugt i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP9:%.*]] = select i1 [[TMP8]], i64 [[TMP7]], i64 0
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = mul i64 [[TMP10]], 8
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT:%.*]] = add i64 0, [[TMP11]]
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP13:%.*]] = mul i64 [[TMP12]], 16
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT1:%.*]] = add i64 0, [[TMP13]]
; CHECK-ORDERED-TF-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = mul i64 [[TMP14]], 24
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT2:%.*]] = add i64 0, [[TMP15]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 0, i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY3:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY4:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT1]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY5:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT2]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED-TF: vector.body:
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK6:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY3]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT12:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK7:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY4]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT13:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK8:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY5]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT14:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP53:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP16]]
; CHECK-ORDERED-TF-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i32 0
; CHECK-ORDERED-TF-NEXT: [[TMP37:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP38:%.*]] = mul i64 [[TMP37]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP39:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP38]]
; CHECK-ORDERED-TF-NEXT: [[TMP40:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP41:%.*]] = mul i64 [[TMP40]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP42:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP41]]
; CHECK-ORDERED-TF-NEXT: [[TMP43:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP44:%.*]] = mul i64 [[TMP43]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP45:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP44]]
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP36]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD9:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP39]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD10:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP42]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD11:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP45]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP46:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> [[WIDE_MASKED_LOAD]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP47:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[TMP46]])
; CHECK-ORDERED-TF-NEXT: [[TMP48:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> [[WIDE_MASKED_LOAD9]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP49:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP47]], <vscale x 8 x float> [[TMP48]])
; CHECK-ORDERED-TF-NEXT: [[TMP50:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> [[WIDE_MASKED_LOAD10]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP51:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP49]], <vscale x 8 x float> [[TMP50]])
; CHECK-ORDERED-TF-NEXT: [[TMP52:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> [[WIDE_MASKED_LOAD11]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP53]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP51]], <vscale x 8 x float> [[TMP52]])
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
; CHECK-ORDERED-TF-NEXT: [[TMP54:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP55:%.*]] = mul i64 [[TMP54]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP56:%.*]] = add i64 [[INDEX]], [[TMP55]]
; CHECK-ORDERED-TF-NEXT: [[TMP57:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP58:%.*]] = mul i64 [[TMP57]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP59:%.*]] = add i64 [[INDEX]], [[TMP58]]
; CHECK-ORDERED-TF-NEXT: [[TMP60:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP61:%.*]] = mul i64 [[TMP60]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP62:%.*]] = add i64 [[INDEX]], [[TMP61]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT12]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP56]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT13]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP59]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT14]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP62]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[TMP63:%.*]] = xor <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 8 x i1> insertelement (<vscale x 8 x i1> poison, i1 true, i64 0), <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP64:%.*]] = extractelement <vscale x 8 x i1> [[TMP63]], i32 0
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP64]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK-ORDERED-TF: middle.block:
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED-TF: scalar.ph:
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP53]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP65:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-TF-NEXT: [[ADD]] = fadd float [[TMP65]], [[SUM_07]]
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[ADD_LCSSA:%.*]] = phi float [ [[ADD]], [[FOR_BODY]] ], [ [[TMP53]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: ret float [[ADD_LCSSA]]
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%sum.07 = phi float [ 0.000000e+00, %entry ], [ %add, %for.body ]
%arrayidx = getelementptr inbounds float, ptr %a, i64 %iv
%0 = load float, ptr %arrayidx, align 4
%add = fadd float %0, %sum.07
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !1
for.end:
ret float %add
}
define void @fadd_strict_interleave(ptr noalias nocapture readonly %a, ptr noalias nocapture readonly %b, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define void @fadd_strict_interleave
; CHECK-NOT-VECTORIZED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDXA:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-NOT-VECTORIZED-NEXT: [[A1:%.*]] = load float, ptr [[A]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[A2:%.*]] = load float, ptr [[ARRAYIDXA]], align 4
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[ADD_PHI1:%.*]] = phi float [ [[A2]], [[ENTRY:%.*]] ], [ [[ADD2:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ADD_PHI2:%.*]] = phi float [ [[A1]], [[ENTRY]] ], [ [[ADD1:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDXB1:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDXB1]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ADD1]] = fadd float [[TMP0]], [[ADD_PHI2]]
; CHECK-NOT-VECTORIZED-NEXT: [[OR:%.*]] = or disjoint i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDXB2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[OR]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDXB2]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ADD2]] = fadd float [[TMP1]], [[ADD_PHI1]]
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 2
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[ADD1_LCSSA:%.*]] = phi float [ [[ADD1]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ADD2_LCSSA:%.*]] = phi float [ [[ADD2]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: store float [[ADD1_LCSSA]], ptr [[A]], align 4
; CHECK-NOT-VECTORIZED-NEXT: store float [[ADD2_LCSSA]], ptr [[ARRAYIDXA]], align 4
; CHECK-NOT-VECTORIZED-NEXT: ret void
;
; CHECK-UNORDERED-LABEL: define void @fadd_strict_interleave
; CHECK-UNORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[ARRAYIDXA:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-UNORDERED-NEXT: [[A1:%.*]] = load float, ptr [[A]], align 4
; CHECK-UNORDERED-NEXT: [[A2:%.*]] = load float, ptr [[ARRAYIDXA]], align 4
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = add i64 [[N]], -2
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = lshr i64 [[TMP0]], 1
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = add nuw i64 [[TMP1]], 1
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 4
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP2]], [[TMP4]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 4
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP2]], [[TMP6]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP2]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[IND_END:%.*]] = mul i64 [[N_VEC]], 2
; CHECK-UNORDERED-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP8:%.*]] = mul i64 [[TMP7]], 4
; CHECK-UNORDERED-NEXT: [[TMP9:%.*]] = insertelement <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), float [[A2]], i32 0
; CHECK-UNORDERED-NEXT: [[TMP10:%.*]] = insertelement <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), float [[A1]], i32 0
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x float> [ [[TMP9]], [[VECTOR_PH]] ], [ [[TMP17:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 4 x float> [ [[TMP10]], [[VECTOR_PH]] ], [ [[TMP16:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
; CHECK-UNORDERED-NEXT: [[TMP11:%.*]] = add i64 [[OFFSET_IDX]], 0
; CHECK-UNORDERED-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP11]]
; CHECK-UNORDERED-NEXT: [[TMP13:%.*]] = getelementptr inbounds float, ptr [[TMP12]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x float>, ptr [[TMP13]], align 4
; CHECK-UNORDERED-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave2.nxv8f32(<vscale x 8 x float> [[WIDE_VEC]])
; CHECK-UNORDERED-NEXT: [[TMP14:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 0
; CHECK-UNORDERED-NEXT: [[TMP15:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 1
; CHECK-UNORDERED-NEXT: [[TMP16]] = fadd <vscale x 4 x float> [[TMP14]], [[VEC_PHI1]]
; CHECK-UNORDERED-NEXT: [[TMP17]] = fadd <vscale x 4 x float> [[TMP15]], [[VEC_PHI]]
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-UNORDERED-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[TMP19:%.*]] = call float @llvm.vector.reduce.fadd.nxv4f32(float -0.000000e+00, <vscale x 4 x float> [[TMP17]])
; CHECK-UNORDERED-NEXT: [[TMP20:%.*]] = call float @llvm.vector.reduce.fadd.nxv4f32(float -0.000000e+00, <vscale x 4 x float> [[TMP16]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP19]], [[MIDDLE_BLOCK]] ], [ [[A2]], [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX2:%.*]] = phi float [ [[TMP20]], [[MIDDLE_BLOCK]] ], [ [[A1]], [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[ADD_PHI1:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD2:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[ADD_PHI2:%.*]] = phi float [ [[BC_MERGE_RDX2]], [[SCALAR_PH]] ], [ [[ADD1:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDXB1:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP21:%.*]] = load float, ptr [[ARRAYIDXB1]], align 4
; CHECK-UNORDERED-NEXT: [[ADD1]] = fadd float [[TMP21]], [[ADD_PHI2]]
; CHECK-UNORDERED-NEXT: [[OR:%.*]] = or disjoint i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[ARRAYIDXB2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[OR]]
; CHECK-UNORDERED-NEXT: [[TMP22:%.*]] = load float, ptr [[ARRAYIDXB2]], align 4
; CHECK-UNORDERED-NEXT: [[ADD2]] = fadd float [[TMP22]], [[ADD_PHI1]]
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 2
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[ADD1_LCSSA:%.*]] = phi float [ [[ADD1]], [[FOR_BODY]] ], [ [[TMP20]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: [[ADD2_LCSSA:%.*]] = phi float [ [[ADD2]], [[FOR_BODY]] ], [ [[TMP19]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: store float [[ADD1_LCSSA]], ptr [[A]], align 4
; CHECK-UNORDERED-NEXT: store float [[ADD2_LCSSA]], ptr [[ARRAYIDXA]], align 4
; CHECK-UNORDERED-NEXT: ret void
;
; CHECK-ORDERED-LABEL: define void @fadd_strict_interleave
; CHECK-ORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: [[ARRAYIDXA:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-ORDERED-NEXT: [[A1:%.*]] = load float, ptr [[A]], align 4
; CHECK-ORDERED-NEXT: [[A2:%.*]] = load float, ptr [[ARRAYIDXA]], align 4
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = add i64 [[N]], -2
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = lshr i64 [[TMP0]], 1
; CHECK-ORDERED-NEXT: [[TMP2:%.*]] = add nuw i64 [[TMP1]], 1
; CHECK-ORDERED-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 4
; CHECK-ORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP2]], [[TMP4]]
; CHECK-ORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED: vector.ph:
; CHECK-ORDERED-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 4
; CHECK-ORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP2]], [[TMP6]]
; CHECK-ORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP2]], [[N_MOD_VF]]
; CHECK-ORDERED-NEXT: [[IND_END:%.*]] = mul i64 [[N_VEC]], 2
; CHECK-ORDERED-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP8:%.*]] = mul i64 [[TMP7]], 4
; CHECK-ORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED: vector.body:
; CHECK-ORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI:%.*]] = phi float [ [[A2]], [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI1:%.*]] = phi float [ [[A1]], [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
; CHECK-ORDERED-NEXT: [[TMP9:%.*]] = add i64 [[OFFSET_IDX]], 0
; CHECK-ORDERED-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP9]]
; CHECK-ORDERED-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[TMP10]], i32 0
; CHECK-ORDERED-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x float>, ptr [[TMP11]], align 4
; CHECK-ORDERED-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave2.nxv8f32(<vscale x 8 x float> [[WIDE_VEC]])
; CHECK-ORDERED-NEXT: [[TMP12:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 0
; CHECK-ORDERED-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 1
; CHECK-ORDERED-NEXT: [[TMP14]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP13]])
; CHECK-ORDERED-NEXT: [[TMP15]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI1]], <vscale x 4 x float> [[TMP12]])
; CHECK-ORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-ORDERED-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK-ORDERED: middle.block:
; CHECK-ORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED: scalar.ph:
; CHECK-ORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP14]], [[MIDDLE_BLOCK]] ], [ [[A2]], [[ENTRY]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX2:%.*]] = phi float [ [[TMP15]], [[MIDDLE_BLOCK]] ], [ [[A1]], [[ENTRY]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[ADD_PHI1:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD2:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[ADD_PHI2:%.*]] = phi float [ [[BC_MERGE_RDX2]], [[SCALAR_PH]] ], [ [[ADD1:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDXB1:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP17:%.*]] = load float, ptr [[ARRAYIDXB1]], align 4
; CHECK-ORDERED-NEXT: [[ADD1]] = fadd float [[TMP17]], [[ADD_PHI2]]
; CHECK-ORDERED-NEXT: [[OR:%.*]] = or disjoint i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[ARRAYIDXB2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[OR]]
; CHECK-ORDERED-NEXT: [[TMP18:%.*]] = load float, ptr [[ARRAYIDXB2]], align 4
; CHECK-ORDERED-NEXT: [[ADD2]] = fadd float [[TMP18]], [[ADD_PHI1]]
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 2
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[ADD1_LCSSA:%.*]] = phi float [ [[ADD1]], [[FOR_BODY]] ], [ [[TMP15]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: [[ADD2_LCSSA:%.*]] = phi float [ [[ADD2]], [[FOR_BODY]] ], [ [[TMP14]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: store float [[ADD1_LCSSA]], ptr [[A]], align 4
; CHECK-ORDERED-NEXT: store float [[ADD2_LCSSA]], ptr [[ARRAYIDXA]], align 4
; CHECK-ORDERED-NEXT: ret void
;
; CHECK-ORDERED-TF-LABEL: define void @fadd_strict_interleave
; CHECK-ORDERED-TF-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDXA:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-ORDERED-TF-NEXT: [[A1:%.*]] = load float, ptr [[A]], align 4
; CHECK-ORDERED-TF-NEXT: [[A2:%.*]] = load float, ptr [[ARRAYIDXA]], align 4
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = add i64 [[N]], -2
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = lshr i64 [[TMP0]], 1
; CHECK-ORDERED-TF-NEXT: [[TMP2:%.*]] = add nuw i64 [[TMP1]], 1
; CHECK-ORDERED-TF-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED-TF: vector.ph:
; CHECK-ORDERED-TF-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP5:%.*]] = sub i64 [[TMP4]], 1
; CHECK-ORDERED-TF-NEXT: [[N_RND_UP:%.*]] = add i64 [[TMP2]], [[TMP5]]
; CHECK-ORDERED-TF-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP4]]
; CHECK-ORDERED-TF-NEXT: [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
; CHECK-ORDERED-TF-NEXT: [[IND_END:%.*]] = mul i64 [[N_VEC]], 2
; CHECK-ORDERED-TF-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP7:%.*]] = mul i64 [[TMP6]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP9:%.*]] = mul i64 [[TMP8]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = sub i64 [[TMP2]], [[TMP9]]
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = icmp ugt i64 [[TMP2]], [[TMP9]]
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = select i1 [[TMP11]], i64 [[TMP10]], i64 0
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 0, i64 [[TMP2]])
; CHECK-ORDERED-TF-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED-TF: vector.body:
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ [[A2]], [[VECTOR_PH]] ], [ [[TMP19:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI1:%.*]] = phi float [ [[A1]], [[VECTOR_PH]] ], [ [[TMP21:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
; CHECK-ORDERED-TF-NEXT: [[TMP13:%.*]] = add i64 [[OFFSET_IDX]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP13]]
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = getelementptr inbounds float, ptr [[TMP14]], i32 0
; CHECK-ORDERED-TF-NEXT: [[INTERLEAVED_MASK:%.*]] = call <vscale x 8 x i1> @llvm.vector.interleave2.nxv8i1(<vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_VEC:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP15]], i32 4, <vscale x 8 x i1> [[INTERLEAVED_MASK]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave2.nxv8f32(<vscale x 8 x float> [[WIDE_MASKED_VEC]])
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP17:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 1
; CHECK-ORDERED-TF-NEXT: [[TMP18:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[TMP17]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP19]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP18]])
; CHECK-ORDERED-TF-NEXT: [[TMP20:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[TMP16]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP21]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI1]], <vscale x 4 x float> [[TMP20]])
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP7]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP12]])
; CHECK-ORDERED-TF-NEXT: [[TMP22:%.*]] = xor <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP23:%.*]] = extractelement <vscale x 4 x i1> [[TMP22]], i32 0
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK-ORDERED-TF: middle.block:
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED-TF: scalar.ph:
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP19]], [[MIDDLE_BLOCK]] ], [ [[A2]], [[ENTRY]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX2:%.*]] = phi float [ [[TMP21]], [[MIDDLE_BLOCK]] ], [ [[A1]], [[ENTRY]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[ADD_PHI1:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD2:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ADD_PHI2:%.*]] = phi float [ [[BC_MERGE_RDX2]], [[SCALAR_PH]] ], [ [[ADD1:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDXB1:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP24:%.*]] = load float, ptr [[ARRAYIDXB1]], align 4
; CHECK-ORDERED-TF-NEXT: [[ADD1]] = fadd float [[TMP24]], [[ADD_PHI2]]
; CHECK-ORDERED-TF-NEXT: [[OR:%.*]] = or disjoint i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDXB2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[OR]]
; CHECK-ORDERED-TF-NEXT: [[TMP25:%.*]] = load float, ptr [[ARRAYIDXB2]], align 4
; CHECK-ORDERED-TF-NEXT: [[ADD2]] = fadd float [[TMP25]], [[ADD_PHI1]]
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 2
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[ADD1_LCSSA:%.*]] = phi float [ [[ADD1]], [[FOR_BODY]] ], [ [[TMP21]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: [[ADD2_LCSSA:%.*]] = phi float [ [[ADD2]], [[FOR_BODY]] ], [ [[TMP19]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: store float [[ADD1_LCSSA]], ptr [[A]], align 4
; CHECK-ORDERED-TF-NEXT: store float [[ADD2_LCSSA]], ptr [[ARRAYIDXA]], align 4
; CHECK-ORDERED-TF-NEXT: ret void
;
entry:
%arrayidxa = getelementptr inbounds float, ptr %a, i64 1
%a1 = load float, ptr %a, align 4
%a2 = load float, ptr %arrayidxa, align 4
br label %for.body
for.body:
%add.phi1 = phi float [ %a2, %entry ], [ %add2, %for.body ]
%add.phi2 = phi float [ %a1, %entry ], [ %add1, %for.body ]
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidxb1 = getelementptr inbounds float, ptr %b, i64 %iv
%0 = load float, ptr %arrayidxb1, align 4
%add1 = fadd float %0, %add.phi2
%or = or disjoint i64 %iv, 1
%arrayidxb2 = getelementptr inbounds float, ptr %b, i64 %or
%1 = load float, ptr %arrayidxb2, align 4
%add2 = fadd float %1, %add.phi1
%iv.next = add nuw nsw i64 %iv, 2
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !2
for.end:
store float %add1, ptr %a, align 4
store float %add2, ptr %arrayidxa, align 4
ret void
}
define float @fadd_of_sum(ptr noalias nocapture readonly %a, ptr noalias nocapture readonly %b, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define float @fadd_of_sum
; CHECK-NOT-VECTORIZED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[CMP1:%.*]] = fcmp ogt float [[TMP0]], 5.000000e-01
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[CMP1]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; CHECK-NOT-VECTORIZED: for.body.preheader:
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[RES_014:%.*]] = phi float [ [[RDX:%.*]], [[FOR_BODY]] ], [ 0.000000e+00, [[FOR_BODY_PREHEADER]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP2:%.*]] = load float, ptr [[ARRAYIDX4]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[TMP2]]
; CHECK-NOT-VECTORIZED-NEXT: [[RDX]] = fadd float [[RES_014]], [[ADD]]
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP6]]
; CHECK-NOT-VECTORIZED: for.end.loopexit:
; CHECK-NOT-VECTORIZED-NEXT: [[RDX_LCSSA:%.*]] = phi float [ [[RDX]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_END]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[RES:%.*]] = phi float [ 0.000000e+00, [[ENTRY:%.*]] ], [ [[RDX_LCSSA]], [[FOR_END_LOOPEXIT]] ]
; CHECK-NOT-VECTORIZED-NEXT: ret float [[RES]]
;
; CHECK-UNORDERED-LABEL: define float @fadd_of_sum
; CHECK-UNORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-UNORDERED-NEXT: [[CMP1:%.*]] = fcmp ogt float [[TMP0]], 5.000000e-01
; CHECK-UNORDERED-NEXT: br i1 [[CMP1]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; CHECK-UNORDERED: for.body.preheader:
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = mul i64 [[TMP1]], 4
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP2]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 4
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP4]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 4
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x float> [ insertelement (<vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), float 0.000000e+00, i32 0), [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[TMP7:%.*]] = add i64 [[INDEX]], 0
; CHECK-UNORDERED-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP7]]
; CHECK-UNORDERED-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[TMP8]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP9]], align 4
; CHECK-UNORDERED-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP7]]
; CHECK-UNORDERED-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[TMP10]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 4 x float>, ptr [[TMP11]], align 4
; CHECK-UNORDERED-NEXT: [[TMP12:%.*]] = fadd <vscale x 4 x float> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-UNORDERED-NEXT: [[TMP13]] = fadd <vscale x 4 x float> [[VEC_PHI]], [[TMP12]]
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[TMP15:%.*]] = call float @llvm.vector.reduce.fadd.nxv4f32(float -0.000000e+00, <vscale x 4 x float> [[TMP13]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP15]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[FOR_BODY_PREHEADER]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-UNORDERED-NEXT: [[RES_014:%.*]] = phi float [ [[RDX:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP16:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-UNORDERED-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP17:%.*]] = load float, ptr [[ARRAYIDX4]], align 4
; CHECK-UNORDERED-NEXT: [[ADD:%.*]] = fadd float [[TMP16]], [[TMP17]]
; CHECK-UNORDERED-NEXT: [[RDX]] = fadd float [[RES_014]], [[ADD]]
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK-UNORDERED: for.end.loopexit:
; CHECK-UNORDERED-NEXT: [[RDX_LCSSA:%.*]] = phi float [ [[RDX]], [[FOR_BODY]] ], [ [[TMP15]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_END]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[RES:%.*]] = phi float [ 0.000000e+00, [[ENTRY:%.*]] ], [ [[RDX_LCSSA]], [[FOR_END_LOOPEXIT]] ]
; CHECK-UNORDERED-NEXT: ret float [[RES]]
;
; CHECK-ORDERED-LABEL: define float @fadd_of_sum
; CHECK-ORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-NEXT: [[CMP1:%.*]] = fcmp ogt float [[TMP0]], 5.000000e-01
; CHECK-ORDERED-NEXT: br i1 [[CMP1]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; CHECK-ORDERED: for.body.preheader:
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP2:%.*]] = mul i64 [[TMP1]], 4
; CHECK-ORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP2]]
; CHECK-ORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED: vector.ph:
; CHECK-ORDERED-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 4
; CHECK-ORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP4]]
; CHECK-ORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-ORDERED-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 4
; CHECK-ORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED: vector.body:
; CHECK-ORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[TMP7:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP7]]
; CHECK-ORDERED-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[TMP8]], i32 0
; CHECK-ORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP9]], align 4
; CHECK-ORDERED-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP7]]
; CHECK-ORDERED-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[TMP10]], i32 0
; CHECK-ORDERED-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 4 x float>, ptr [[TMP11]], align 4
; CHECK-ORDERED-NEXT: [[TMP12:%.*]] = fadd <vscale x 4 x float> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-ORDERED-NEXT: [[TMP13]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP12]])
; CHECK-ORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK-ORDERED: middle.block:
; CHECK-ORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED: scalar.ph:
; CHECK-ORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP13]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[FOR_BODY_PREHEADER]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-ORDERED-NEXT: [[RES_014:%.*]] = phi float [ [[RDX:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP15:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP16:%.*]] = load float, ptr [[ARRAYIDX4]], align 4
; CHECK-ORDERED-NEXT: [[ADD:%.*]] = fadd float [[TMP15]], [[TMP16]]
; CHECK-ORDERED-NEXT: [[RDX]] = fadd float [[RES_014]], [[ADD]]
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK-ORDERED: for.end.loopexit:
; CHECK-ORDERED-NEXT: [[RDX_LCSSA:%.*]] = phi float [ [[RDX]], [[FOR_BODY]] ], [ [[TMP13]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_END]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[RES:%.*]] = phi float [ 0.000000e+00, [[ENTRY:%.*]] ], [ [[RDX_LCSSA]], [[FOR_END_LOOPEXIT]] ]
; CHECK-ORDERED-NEXT: ret float [[RES]]
;
; CHECK-ORDERED-TF-LABEL: define float @fadd_of_sum
; CHECK-ORDERED-TF-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 1
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-TF-NEXT: [[CMP1:%.*]] = fcmp ogt float [[TMP0]], 5.000000e-01
; CHECK-ORDERED-TF-NEXT: br i1 [[CMP1]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; CHECK-ORDERED-TF: for.body.preheader:
; CHECK-ORDERED-TF-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED-TF: vector.ph:
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP2:%.*]] = mul i64 [[TMP1]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP3:%.*]] = sub i64 [[TMP2]], 1
; CHECK-ORDERED-TF-NEXT: [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP3]]
; CHECK-ORDERED-TF-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP2]]
; CHECK-ORDERED-TF-NEXT: [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
; CHECK-ORDERED-TF-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP7:%.*]] = mul i64 [[TMP6]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP8:%.*]] = sub i64 [[N]], [[TMP7]]
; CHECK-ORDERED-TF-NEXT: [[TMP9:%.*]] = icmp ugt i64 [[N]], [[TMP7]]
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = select i1 [[TMP9]], i64 [[TMP8]], i64 0
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 0, i64 [[N]])
; CHECK-ORDERED-TF-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED-TF: vector.body:
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP11]]
; CHECK-ORDERED-TF-NEXT: [[TMP13:%.*]] = getelementptr inbounds float, ptr [[TMP12]], i32 0
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP13]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP11]]
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = getelementptr inbounds float, ptr [[TMP14]], i32 0
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP15]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = fadd <vscale x 4 x float> [[WIDE_MASKED_LOAD]], [[WIDE_MASKED_LOAD1]]
; CHECK-ORDERED-TF-NEXT: [[TMP17:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[TMP16]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP18]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP17]])
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP5]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP10]])
; CHECK-ORDERED-TF-NEXT: [[TMP19:%.*]] = xor <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP20:%.*]] = extractelement <vscale x 4 x i1> [[TMP19]], i32 0
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK-ORDERED-TF: middle.block:
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED-TF: scalar.ph:
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP18]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[FOR_BODY_PREHEADER]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-ORDERED-TF-NEXT: [[RES_014:%.*]] = phi float [ [[RDX:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP21:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP22:%.*]] = load float, ptr [[ARRAYIDX4]], align 4
; CHECK-ORDERED-TF-NEXT: [[ADD:%.*]] = fadd float [[TMP21]], [[TMP22]]
; CHECK-ORDERED-TF-NEXT: [[RDX]] = fadd float [[RES_014]], [[ADD]]
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK-ORDERED-TF: for.end.loopexit:
; CHECK-ORDERED-TF-NEXT: [[RDX_LCSSA:%.*]] = phi float [ [[RDX]], [[FOR_BODY]] ], [ [[TMP18]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_END]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[RES:%.*]] = phi float [ 0.000000e+00, [[ENTRY:%.*]] ], [ [[RDX_LCSSA]], [[FOR_END_LOOPEXIT]] ]
; CHECK-ORDERED-TF-NEXT: ret float [[RES]]
;
entry:
%arrayidx = getelementptr inbounds float, ptr %a, i64 1
%0 = load float, ptr %arrayidx, align 4
%cmp1 = fcmp ogt float %0, 5.000000e-01
br i1 %cmp1, label %for.body, label %for.end
for.body: ; preds = %for.body
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%res.014 = phi float [ 0.000000e+00, %entry ], [ %rdx, %for.body ]
%arrayidx2 = getelementptr inbounds float, ptr %a, i64 %iv
%1 = load float, ptr %arrayidx2, align 4
%arrayidx4 = getelementptr inbounds float, ptr %b, i64 %iv
%2 = load float, ptr %arrayidx4, align 4
%add = fadd float %1, %2
%rdx = fadd float %res.014, %add
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !2
for.end: ; preds = %for.body, %entry
%res = phi float [ 0.000000e+00, %entry ], [ %rdx, %for.body ]
ret float %res
}
define float @fadd_conditional(ptr noalias nocapture readonly %a, ptr noalias nocapture readonly %b, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define float @fadd_conditional
; CHECK-NOT-VECTORIZED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[RES:%.*]] = phi float [ 1.000000e+00, [[ENTRY]] ], [ [[FADD:%.*]], [[FOR_INC]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP0]], 0.000000e+00
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK-NOT-VECTORIZED: if.then:
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_INC]]
; CHECK-NOT-VECTORIZED: for.inc:
; CHECK-NOT-VECTORIZED-NEXT: [[PHI:%.*]] = phi float [ [[TMP1]], [[IF_THEN]] ], [ 3.000000e+00, [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[FADD]] = fadd float [[RES]], [[PHI]]
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP6]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[RDX:%.*]] = phi float [ [[FADD]], [[FOR_INC]] ]
; CHECK-NOT-VECTORIZED-NEXT: ret float [[RDX]]
;
; CHECK-UNORDERED-LABEL: define float @fadd_conditional
; CHECK-UNORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x float> [ insertelement (<vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), float 1.000000e+00, i32 0), [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-UNORDERED-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP7]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP8]], align 4
; CHECK-UNORDERED-NEXT: [[TMP9:%.*]] = fcmp une <vscale x 4 x float> [[WIDE_LOAD]], zeroinitializer
; CHECK-UNORDERED-NEXT: [[TMP10:%.*]] = getelementptr float, ptr [[A]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP11:%.*]] = getelementptr float, ptr [[TMP10]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[TMP9]], <vscale x 4 x float> poison)
; CHECK-UNORDERED-NEXT: [[TMP12:%.*]] = xor <vscale x 4 x i1> [[TMP9]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-UNORDERED-NEXT: [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP12]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x float> [[WIDE_MASKED_LOAD]]
; CHECK-UNORDERED-NEXT: [[TMP13]] = fadd <vscale x 4 x float> [[VEC_PHI]], [[PREDPHI]]
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-UNORDERED-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[TMP15:%.*]] = call float @llvm.vector.reduce.fadd.nxv4f32(float -0.000000e+00, <vscale x 4 x float> [[TMP13]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP15]], [[MIDDLE_BLOCK]] ], [ 1.000000e+00, [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-UNORDERED-NEXT: [[RES:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[FADD:%.*]], [[FOR_INC]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP16:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-UNORDERED-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP16]], 0.000000e+00
; CHECK-UNORDERED-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK-UNORDERED: if.then:
; CHECK-UNORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP17:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-UNORDERED-NEXT: br label [[FOR_INC]]
; CHECK-UNORDERED: for.inc:
; CHECK-UNORDERED-NEXT: [[PHI:%.*]] = phi float [ [[TMP17]], [[IF_THEN]] ], [ 3.000000e+00, [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[FADD]] = fadd float [[RES]], [[PHI]]
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[RDX:%.*]] = phi float [ [[FADD]], [[FOR_INC]] ], [ [[TMP15]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: ret float [[RDX]]
;
; CHECK-ORDERED-LABEL: define float @fadd_conditional
; CHECK-ORDERED-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; CHECK-ORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-ORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED: vector.ph:
; CHECK-ORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; CHECK-ORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-ORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-ORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; CHECK-ORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED: vector.body:
; CHECK-ORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI:%.*]] = phi float [ 1.000000e+00, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP7]], i32 0
; CHECK-ORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP8]], align 4
; CHECK-ORDERED-NEXT: [[TMP9:%.*]] = fcmp une <vscale x 4 x float> [[WIDE_LOAD]], zeroinitializer
; CHECK-ORDERED-NEXT: [[TMP10:%.*]] = getelementptr float, ptr [[A]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP11:%.*]] = getelementptr float, ptr [[TMP10]], i32 0
; CHECK-ORDERED-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[TMP9]], <vscale x 4 x float> poison)
; CHECK-ORDERED-NEXT: [[TMP12:%.*]] = xor <vscale x 4 x i1> [[TMP9]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-NEXT: [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP12]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x float> [[WIDE_MASKED_LOAD]]
; CHECK-ORDERED-NEXT: [[TMP13]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[PREDPHI]])
; CHECK-ORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-ORDERED-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK-ORDERED: middle.block:
; CHECK-ORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED: scalar.ph:
; CHECK-ORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP13]], [[MIDDLE_BLOCK]] ], [ 1.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-ORDERED-NEXT: [[RES:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[FADD:%.*]], [[FOR_INC]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP15:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP15]], 0.000000e+00
; CHECK-ORDERED-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK-ORDERED: if.then:
; CHECK-ORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP16:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-NEXT: br label [[FOR_INC]]
; CHECK-ORDERED: for.inc:
; CHECK-ORDERED-NEXT: [[PHI:%.*]] = phi float [ [[TMP16]], [[IF_THEN]] ], [ 3.000000e+00, [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[FADD]] = fadd float [[RES]], [[PHI]]
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[RDX:%.*]] = phi float [ [[FADD]], [[FOR_INC]] ], [ [[TMP13]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: ret float [[RDX]]
;
; CHECK-ORDERED-TF-LABEL: define float @fadd_conditional
; CHECK-ORDERED-TF-SAME: (ptr noalias nocapture readonly [[A:%.*]], ptr noalias nocapture readonly [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED-TF: vector.ph:
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP2:%.*]] = sub i64 [[TMP1]], 1
; CHECK-ORDERED-TF-NEXT: [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP2]]
; CHECK-ORDERED-TF-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP1]]
; CHECK-ORDERED-TF-NEXT: [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
; CHECK-ORDERED-TF-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 4
; CHECK-ORDERED-TF-NEXT: [[TMP7:%.*]] = sub i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP8:%.*]] = icmp ugt i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP9:%.*]] = select i1 [[TMP8]], i64 [[TMP7]], i64 0
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 0, i64 [[N]])
; CHECK-ORDERED-TF-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED-TF: vector.body:
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 1.000000e+00, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP10]]
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP11]], i32 0
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP12]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP13:%.*]] = fcmp une <vscale x 4 x float> [[WIDE_MASKED_LOAD]], zeroinitializer
; CHECK-ORDERED-TF-NEXT: [[TMP14:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i1> [[TMP13]], <vscale x 4 x i1> zeroinitializer
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = getelementptr float, ptr [[A]], i64 [[TMP10]]
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = getelementptr float, ptr [[TMP15]], i32 0
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP16]], i32 4, <vscale x 4 x i1> [[TMP14]], <vscale x 4 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP14]], <vscale x 4 x float> [[WIDE_MASKED_LOAD1]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP17:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[PREDPHI]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP18]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP17]])
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[TMP19:%.*]] = xor <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP20:%.*]] = extractelement <vscale x 4 x i1> [[TMP19]], i32 0
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK-ORDERED-TF: middle.block:
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED-TF: scalar.ph:
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP18]], [[MIDDLE_BLOCK]] ], [ 1.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-ORDERED-TF-NEXT: [[RES:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[FADD:%.*]], [[FOR_INC]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP21:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-TF-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP21]], 0.000000e+00
; CHECK-ORDERED-TF-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK-ORDERED-TF: if.then:
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP22:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-TF-NEXT: br label [[FOR_INC]]
; CHECK-ORDERED-TF: for.inc:
; CHECK-ORDERED-TF-NEXT: [[PHI:%.*]] = phi float [ [[TMP22]], [[IF_THEN]] ], [ 3.000000e+00, [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[FADD]] = fadd float [[RES]], [[PHI]]
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[RDX:%.*]] = phi float [ [[FADD]], [[FOR_INC]] ], [ [[TMP18]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: ret float [[RDX]]
;
entry:
br label %for.body
for.body: ; preds = %for.body
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.inc ]
%res = phi float [ 1.000000e+00, %entry ], [ %fadd, %for.inc ]
%arrayidx = getelementptr inbounds float, ptr %b, i64 %iv
%0 = load float, ptr %arrayidx, align 4
%tobool = fcmp une float %0, 0.000000e+00
br i1 %tobool, label %if.then, label %for.inc
if.then: ; preds = %for.body
%arrayidx2 = getelementptr inbounds float, ptr %a, i64 %iv
%1 = load float, ptr %arrayidx2, align 4
br label %for.inc
for.inc:
%phi = phi float [ %1, %if.then ], [ 3.000000e+00, %for.body ]
%fadd = fadd float %res, %phi
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !2
for.end:
%rdx = phi float [ %fadd, %for.inc ]
ret float %rdx
}
; Negative test - loop contains multiple fadds which we cannot safely reorder
define float @fadd_multiple(ptr noalias nocapture %a, ptr noalias nocapture %b, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define float @fadd_multiple
; CHECK-NOT-VECTORIZED-SAME: (ptr noalias nocapture [[A:%.*]], ptr noalias nocapture [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[SUM:%.*]] = phi float [ -0.000000e+00, [[ENTRY]] ], [ [[ADD3:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ADD:%.*]] = fadd float [[SUM]], [[TMP0]]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ADD3]] = fadd float [[ADD]], [[TMP1]]
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP0]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[RDX:%.*]] = phi float [ [[ADD3]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: ret float [[RDX]]
;
; CHECK-UNORDERED-LABEL: define float @fadd_multiple
; CHECK-UNORDERED-SAME: (ptr noalias nocapture [[A:%.*]], ptr noalias nocapture [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 8
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 8
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 8
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 8 x float> [ insertelement (<vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), float -0.000000e+00, i32 0), [[VECTOR_PH]] ], [ [[TMP12:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-UNORDERED-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP7]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP8]], align 4
; CHECK-UNORDERED-NEXT: [[TMP9:%.*]] = fadd <vscale x 8 x float> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-UNORDERED-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[TMP10]], i32 0
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 8 x float>, ptr [[TMP11]], align 4
; CHECK-UNORDERED-NEXT: [[TMP12]] = fadd <vscale x 8 x float> [[TMP9]], [[WIDE_LOAD1]]
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-UNORDERED-NEXT: [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[TMP14:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float -0.000000e+00, <vscale x 8 x float> [[TMP12]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP14]], [[MIDDLE_BLOCK]] ], [ -0.000000e+00, [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[SUM:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD3:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP15:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-UNORDERED-NEXT: [[ADD:%.*]] = fadd float [[SUM]], [[TMP15]]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP16:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-UNORDERED-NEXT: [[ADD3]] = fadd float [[ADD]], [[TMP16]]
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[RDX:%.*]] = phi float [ [[ADD3]], [[FOR_BODY]] ], [ [[TMP14]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: ret float [[RDX]]
;
; CHECK-ORDERED-LABEL: define float @fadd_multiple
; CHECK-ORDERED-SAME: (ptr noalias nocapture [[A:%.*]], ptr noalias nocapture [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[SUM:%.*]] = phi float [ -0.000000e+00, [[ENTRY]] ], [ [[ADD3:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-NEXT: [[ADD:%.*]] = fadd float [[SUM]], [[TMP0]]
; CHECK-ORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-NEXT: [[ADD3]] = fadd float [[ADD]], [[TMP1]]
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[RDX:%.*]] = phi float [ [[ADD3]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: ret float [[RDX]]
;
; CHECK-ORDERED-TF-LABEL: define float @fadd_multiple
; CHECK-ORDERED-TF-SAME: (ptr noalias nocapture [[A:%.*]], ptr noalias nocapture [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[SUM:%.*]] = phi float [ -0.000000e+00, [[ENTRY]] ], [ [[ADD3:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-TF-NEXT: [[ADD:%.*]] = fadd float [[SUM]], [[TMP0]]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-TF-NEXT: [[ADD3]] = fadd float [[ADD]], [[TMP1]]
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[RDX:%.*]] = phi float [ [[ADD3]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: ret float [[RDX]]
;
entry:
br label %for.body
for.body: ; preds = %entry, %for.body
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%sum = phi float [ -0.000000e+00, %entry ], [ %add3, %for.body ]
%arrayidx = getelementptr inbounds float, ptr %a, i64 %iv
%0 = load float, ptr %arrayidx, align 4
%add = fadd float %sum, %0
%arrayidx2 = getelementptr inbounds float, ptr %b, i64 %iv
%1 = load float, ptr %arrayidx2, align 4
%add3 = fadd float %add, %1
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !0
for.end: ; preds = %for.body
%rdx = phi float [ %add3, %for.body ]
ret float %rdx
}
; Test case where loop has a call to the llvm.fmuladd intrinsic.
define float @fmuladd_strict(ptr %a, ptr %b, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define float @fmuladd_strict
; CHECK-NOT-VECTORIZED-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[SUM_07:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[MULADD]] = tail call float @llvm.fmuladd.f32(float [[TMP0]], float [[TMP1]], float [[SUM_07]])
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP4]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: ret float [[MULADD_LCSSA]]
;
; CHECK-UNORDERED-LABEL: define float @fmuladd_strict
; CHECK-UNORDERED-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 32
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 32
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 8 x float> [ insertelement (<vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), float 0.000000e+00, i32 0), [[VECTOR_PH]] ], [ [[TMP50:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP51:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI2:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP52:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI3:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP53:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-UNORDERED-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i32 0
; CHECK-UNORDERED-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP28:%.*]] = mul i64 [[TMP27]], 8
; CHECK-UNORDERED-NEXT: [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP28]]
; CHECK-UNORDERED-NEXT: [[TMP30:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP31:%.*]] = mul i64 [[TMP30]], 16
; CHECK-UNORDERED-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP31]]
; CHECK-UNORDERED-NEXT: [[TMP33:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP34:%.*]] = mul i64 [[TMP33]], 24
; CHECK-UNORDERED-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP34]]
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP26]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD4:%.*]] = load <vscale x 8 x float>, ptr [[TMP29]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD5:%.*]] = load <vscale x 8 x float>, ptr [[TMP32]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 8 x float>, ptr [[TMP35]], align 4
; CHECK-UNORDERED-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i32 0
; CHECK-UNORDERED-NEXT: [[TMP41:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP42:%.*]] = mul i64 [[TMP41]], 8
; CHECK-UNORDERED-NEXT: [[TMP43:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP42]]
; CHECK-UNORDERED-NEXT: [[TMP44:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP45:%.*]] = mul i64 [[TMP44]], 16
; CHECK-UNORDERED-NEXT: [[TMP46:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP45]]
; CHECK-UNORDERED-NEXT: [[TMP47:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP48:%.*]] = mul i64 [[TMP47]], 24
; CHECK-UNORDERED-NEXT: [[TMP49:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP48]]
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD7:%.*]] = load <vscale x 8 x float>, ptr [[TMP40]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD8:%.*]] = load <vscale x 8 x float>, ptr [[TMP43]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD9:%.*]] = load <vscale x 8 x float>, ptr [[TMP46]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD10:%.*]] = load <vscale x 8 x float>, ptr [[TMP49]], align 4
; CHECK-UNORDERED-NEXT: [[TMP50]] = call <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD]], <vscale x 8 x float> [[WIDE_LOAD7]], <vscale x 8 x float> [[VEC_PHI]])
; CHECK-UNORDERED-NEXT: [[TMP51]] = call <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD4]], <vscale x 8 x float> [[WIDE_LOAD8]], <vscale x 8 x float> [[VEC_PHI1]])
; CHECK-UNORDERED-NEXT: [[TMP52]] = call <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD5]], <vscale x 8 x float> [[WIDE_LOAD9]], <vscale x 8 x float> [[VEC_PHI2]])
; CHECK-UNORDERED-NEXT: [[TMP53]] = call <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD6]], <vscale x 8 x float> [[WIDE_LOAD10]], <vscale x 8 x float> [[VEC_PHI3]])
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-UNORDERED-NEXT: [[TMP54:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP54]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[BIN_RDX:%.*]] = fadd <vscale x 8 x float> [[TMP51]], [[TMP50]]
; CHECK-UNORDERED-NEXT: [[BIN_RDX11:%.*]] = fadd <vscale x 8 x float> [[TMP52]], [[BIN_RDX]]
; CHECK-UNORDERED-NEXT: [[BIN_RDX12:%.*]] = fadd <vscale x 8 x float> [[TMP53]], [[BIN_RDX11]]
; CHECK-UNORDERED-NEXT: [[TMP55:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float -0.000000e+00, <vscale x 8 x float> [[BIN_RDX12]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP55]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP56:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-UNORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP57:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-UNORDERED-NEXT: [[MULADD]] = tail call float @llvm.fmuladd.f32(float [[TMP56]], float [[TMP57]], float [[SUM_07]])
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP15:![0-9]+]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ], [ [[TMP55]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: ret float [[MULADD_LCSSA]]
;
; CHECK-ORDERED-LABEL: define float @fmuladd_strict
; CHECK-ORDERED-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-ORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-ORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED: vector.ph:
; CHECK-ORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 32
; CHECK-ORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-ORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-ORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 32
; CHECK-ORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED: vector.body:
; CHECK-ORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP57:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i32 0
; CHECK-ORDERED-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP28:%.*]] = mul i64 [[TMP27]], 8
; CHECK-ORDERED-NEXT: [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP28]]
; CHECK-ORDERED-NEXT: [[TMP30:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP31:%.*]] = mul i64 [[TMP30]], 16
; CHECK-ORDERED-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP31]]
; CHECK-ORDERED-NEXT: [[TMP33:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP34:%.*]] = mul i64 [[TMP33]], 24
; CHECK-ORDERED-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP34]]
; CHECK-ORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP26]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 8 x float>, ptr [[TMP29]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 8 x float>, ptr [[TMP32]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD3:%.*]] = load <vscale x 8 x float>, ptr [[TMP35]], align 4
; CHECK-ORDERED-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i32 0
; CHECK-ORDERED-NEXT: [[TMP41:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP42:%.*]] = mul i64 [[TMP41]], 8
; CHECK-ORDERED-NEXT: [[TMP43:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP42]]
; CHECK-ORDERED-NEXT: [[TMP44:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP45:%.*]] = mul i64 [[TMP44]], 16
; CHECK-ORDERED-NEXT: [[TMP46:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP45]]
; CHECK-ORDERED-NEXT: [[TMP47:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP48:%.*]] = mul i64 [[TMP47]], 24
; CHECK-ORDERED-NEXT: [[TMP49:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP48]]
; CHECK-ORDERED-NEXT: [[WIDE_LOAD4:%.*]] = load <vscale x 8 x float>, ptr [[TMP40]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD5:%.*]] = load <vscale x 8 x float>, ptr [[TMP43]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 8 x float>, ptr [[TMP46]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD7:%.*]] = load <vscale x 8 x float>, ptr [[TMP49]], align 4
; CHECK-ORDERED-NEXT: [[TMP50:%.*]] = fmul <vscale x 8 x float> [[WIDE_LOAD]], [[WIDE_LOAD4]]
; CHECK-ORDERED-NEXT: [[TMP51:%.*]] = fmul <vscale x 8 x float> [[WIDE_LOAD1]], [[WIDE_LOAD5]]
; CHECK-ORDERED-NEXT: [[TMP52:%.*]] = fmul <vscale x 8 x float> [[WIDE_LOAD2]], [[WIDE_LOAD6]]
; CHECK-ORDERED-NEXT: [[TMP53:%.*]] = fmul <vscale x 8 x float> [[WIDE_LOAD3]], [[WIDE_LOAD7]]
; CHECK-ORDERED-NEXT: [[TMP54:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[TMP50]])
; CHECK-ORDERED-NEXT: [[TMP55:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP54]], <vscale x 8 x float> [[TMP51]])
; CHECK-ORDERED-NEXT: [[TMP56:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP55]], <vscale x 8 x float> [[TMP52]])
; CHECK-ORDERED-NEXT: [[TMP57]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP56]], <vscale x 8 x float> [[TMP53]])
; CHECK-ORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-ORDERED-NEXT: [[TMP58:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[TMP58]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK-ORDERED: middle.block:
; CHECK-ORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED: scalar.ph:
; CHECK-ORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP57]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP59:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP60:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-NEXT: [[MULADD]] = tail call float @llvm.fmuladd.f32(float [[TMP59]], float [[TMP60]], float [[SUM_07]])
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ], [ [[TMP57]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: ret float [[MULADD_LCSSA]]
;
; CHECK-ORDERED-TF-LABEL: define float @fmuladd_strict
; CHECK-ORDERED-TF-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED-TF: vector.ph:
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP2:%.*]] = sub i64 [[TMP1]], 1
; CHECK-ORDERED-TF-NEXT: [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP2]]
; CHECK-ORDERED-TF-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP1]]
; CHECK-ORDERED-TF-NEXT: [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
; CHECK-ORDERED-TF-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP7:%.*]] = sub i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP8:%.*]] = icmp ugt i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP9:%.*]] = select i1 [[TMP8]], i64 [[TMP7]], i64 0
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = mul i64 [[TMP10]], 8
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT:%.*]] = add i64 0, [[TMP11]]
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP13:%.*]] = mul i64 [[TMP12]], 16
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT1:%.*]] = add i64 0, [[TMP13]]
; CHECK-ORDERED-TF-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = mul i64 [[TMP14]], 24
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT2:%.*]] = add i64 0, [[TMP15]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 0, i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY3:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY4:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT1]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY5:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT2]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED-TF: vector.body:
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK6:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY3]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT16:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK7:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY4]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT17:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK8:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY5]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT18:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP71:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP16]]
; CHECK-ORDERED-TF-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i32 0
; CHECK-ORDERED-TF-NEXT: [[TMP37:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP38:%.*]] = mul i64 [[TMP37]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP39:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP38]]
; CHECK-ORDERED-TF-NEXT: [[TMP40:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP41:%.*]] = mul i64 [[TMP40]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP42:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP41]]
; CHECK-ORDERED-TF-NEXT: [[TMP43:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP44:%.*]] = mul i64 [[TMP43]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP45:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP44]]
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP36]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD9:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP39]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD10:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP42]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD11:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP45]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP46:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP16]]
; CHECK-ORDERED-TF-NEXT: [[TMP50:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i32 0
; CHECK-ORDERED-TF-NEXT: [[TMP51:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP52:%.*]] = mul i64 [[TMP51]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP53:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i64 [[TMP52]]
; CHECK-ORDERED-TF-NEXT: [[TMP54:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP55:%.*]] = mul i64 [[TMP54]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP56:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i64 [[TMP55]]
; CHECK-ORDERED-TF-NEXT: [[TMP57:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP58:%.*]] = mul i64 [[TMP57]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP59:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i64 [[TMP58]]
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD12:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP50]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD13:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP53]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD14:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP56]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD15:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP59]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP60:%.*]] = fmul <vscale x 8 x float> [[WIDE_MASKED_LOAD]], [[WIDE_MASKED_LOAD12]]
; CHECK-ORDERED-TF-NEXT: [[TMP61:%.*]] = fmul <vscale x 8 x float> [[WIDE_MASKED_LOAD9]], [[WIDE_MASKED_LOAD13]]
; CHECK-ORDERED-TF-NEXT: [[TMP62:%.*]] = fmul <vscale x 8 x float> [[WIDE_MASKED_LOAD10]], [[WIDE_MASKED_LOAD14]]
; CHECK-ORDERED-TF-NEXT: [[TMP63:%.*]] = fmul <vscale x 8 x float> [[WIDE_MASKED_LOAD11]], [[WIDE_MASKED_LOAD15]]
; CHECK-ORDERED-TF-NEXT: [[TMP64:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> [[TMP60]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP65:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[TMP64]])
; CHECK-ORDERED-TF-NEXT: [[TMP66:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> [[TMP61]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP67:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP65]], <vscale x 8 x float> [[TMP66]])
; CHECK-ORDERED-TF-NEXT: [[TMP68:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> [[TMP62]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP69:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP67]], <vscale x 8 x float> [[TMP68]])
; CHECK-ORDERED-TF-NEXT: [[TMP70:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> [[TMP63]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP71]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP69]], <vscale x 8 x float> [[TMP70]])
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
; CHECK-ORDERED-TF-NEXT: [[TMP72:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP73:%.*]] = mul i64 [[TMP72]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP74:%.*]] = add i64 [[INDEX]], [[TMP73]]
; CHECK-ORDERED-TF-NEXT: [[TMP75:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP76:%.*]] = mul i64 [[TMP75]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP77:%.*]] = add i64 [[INDEX]], [[TMP76]]
; CHECK-ORDERED-TF-NEXT: [[TMP78:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP79:%.*]] = mul i64 [[TMP78]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP80:%.*]] = add i64 [[INDEX]], [[TMP79]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT16]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP74]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT17]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP77]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT18]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP80]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[TMP81:%.*]] = xor <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 8 x i1> insertelement (<vscale x 8 x i1> poison, i1 true, i64 0), <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP82:%.*]] = extractelement <vscale x 8 x i1> [[TMP81]], i32 0
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP82]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK-ORDERED-TF: middle.block:
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED-TF: scalar.ph:
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP71]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP83:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP84:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-TF-NEXT: [[MULADD]] = tail call float @llvm.fmuladd.f32(float [[TMP83]], float [[TMP84]], float [[SUM_07]])
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ], [ [[TMP71]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: ret float [[MULADD_LCSSA]]
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%sum.07 = phi float [ 0.000000e+00, %entry ], [ %muladd, %for.body ]
%arrayidx = getelementptr inbounds float, ptr %a, i64 %iv
%0 = load float, ptr %arrayidx, align 4
%arrayidx2 = getelementptr inbounds float, ptr %b, i64 %iv
%1 = load float, ptr %arrayidx2, align 4
%muladd = tail call float @llvm.fmuladd.f32(float %0, float %1, float %sum.07)
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !1
for.end:
ret float %muladd
}
; Same as above but where the call to the llvm.fmuladd intrinsic has a fast-math flag.
define float @fmuladd_strict_fmf(ptr %a, ptr %b, i64 %n) #0 {
; CHECK-NOT-VECTORIZED-LABEL: define float @fmuladd_strict_fmf
; CHECK-NOT-VECTORIZED-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NOT-VECTORIZED-NEXT: entry:
; CHECK-NOT-VECTORIZED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-NOT-VECTORIZED: for.body:
; CHECK-NOT-VECTORIZED-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[SUM_07:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-NOT-VECTORIZED-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NOT-VECTORIZED-NEXT: [[MULADD]] = tail call nnan float @llvm.fmuladd.f32(float [[TMP0]], float [[TMP1]], float [[SUM_07]])
; CHECK-NOT-VECTORIZED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NOT-VECTORIZED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NOT-VECTORIZED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP4]]
; CHECK-NOT-VECTORIZED: for.end:
; CHECK-NOT-VECTORIZED-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ]
; CHECK-NOT-VECTORIZED-NEXT: ret float [[MULADD_LCSSA]]
;
; CHECK-UNORDERED-LABEL: define float @fmuladd_strict_fmf
; CHECK-UNORDERED-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-UNORDERED-NEXT: entry:
; CHECK-UNORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-UNORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-UNORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-UNORDERED: vector.ph:
; CHECK-UNORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 32
; CHECK-UNORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-UNORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-UNORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 32
; CHECK-UNORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-UNORDERED: vector.body:
; CHECK-UNORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 8 x float> [ insertelement (<vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), float 0.000000e+00, i32 0), [[VECTOR_PH]] ], [ [[TMP50:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP51:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI2:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP52:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[VEC_PHI3:%.*]] = phi <vscale x 8 x float> [ shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer), [[VECTOR_PH]] ], [ [[TMP53:%.*]], [[VECTOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-UNORDERED-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i32 0
; CHECK-UNORDERED-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP28:%.*]] = mul i64 [[TMP27]], 8
; CHECK-UNORDERED-NEXT: [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP28]]
; CHECK-UNORDERED-NEXT: [[TMP30:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP31:%.*]] = mul i64 [[TMP30]], 16
; CHECK-UNORDERED-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP31]]
; CHECK-UNORDERED-NEXT: [[TMP33:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP34:%.*]] = mul i64 [[TMP33]], 24
; CHECK-UNORDERED-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP34]]
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP26]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD4:%.*]] = load <vscale x 8 x float>, ptr [[TMP29]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD5:%.*]] = load <vscale x 8 x float>, ptr [[TMP32]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 8 x float>, ptr [[TMP35]], align 4
; CHECK-UNORDERED-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP6]]
; CHECK-UNORDERED-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i32 0
; CHECK-UNORDERED-NEXT: [[TMP41:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP42:%.*]] = mul i64 [[TMP41]], 8
; CHECK-UNORDERED-NEXT: [[TMP43:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP42]]
; CHECK-UNORDERED-NEXT: [[TMP44:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP45:%.*]] = mul i64 [[TMP44]], 16
; CHECK-UNORDERED-NEXT: [[TMP46:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP45]]
; CHECK-UNORDERED-NEXT: [[TMP47:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-UNORDERED-NEXT: [[TMP48:%.*]] = mul i64 [[TMP47]], 24
; CHECK-UNORDERED-NEXT: [[TMP49:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP48]]
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD7:%.*]] = load <vscale x 8 x float>, ptr [[TMP40]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD8:%.*]] = load <vscale x 8 x float>, ptr [[TMP43]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD9:%.*]] = load <vscale x 8 x float>, ptr [[TMP46]], align 4
; CHECK-UNORDERED-NEXT: [[WIDE_LOAD10:%.*]] = load <vscale x 8 x float>, ptr [[TMP49]], align 4
; CHECK-UNORDERED-NEXT: [[TMP50]] = call nnan <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD]], <vscale x 8 x float> [[WIDE_LOAD7]], <vscale x 8 x float> [[VEC_PHI]])
; CHECK-UNORDERED-NEXT: [[TMP51]] = call nnan <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD4]], <vscale x 8 x float> [[WIDE_LOAD8]], <vscale x 8 x float> [[VEC_PHI1]])
; CHECK-UNORDERED-NEXT: [[TMP52]] = call nnan <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD5]], <vscale x 8 x float> [[WIDE_LOAD9]], <vscale x 8 x float> [[VEC_PHI2]])
; CHECK-UNORDERED-NEXT: [[TMP53]] = call nnan <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> [[WIDE_LOAD6]], <vscale x 8 x float> [[WIDE_LOAD10]], <vscale x 8 x float> [[VEC_PHI3]])
; CHECK-UNORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-UNORDERED-NEXT: [[TMP54:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[TMP54]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK-UNORDERED: middle.block:
; CHECK-UNORDERED-NEXT: [[BIN_RDX:%.*]] = fadd nnan <vscale x 8 x float> [[TMP51]], [[TMP50]]
; CHECK-UNORDERED-NEXT: [[BIN_RDX11:%.*]] = fadd nnan <vscale x 8 x float> [[TMP52]], [[BIN_RDX]]
; CHECK-UNORDERED-NEXT: [[BIN_RDX12:%.*]] = fadd nnan <vscale x 8 x float> [[TMP53]], [[BIN_RDX11]]
; CHECK-UNORDERED-NEXT: [[TMP55:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float -0.000000e+00, <vscale x 8 x float> [[BIN_RDX12]])
; CHECK-UNORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-UNORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-UNORDERED: scalar.ph:
; CHECK-UNORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-UNORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP55]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-UNORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-UNORDERED: for.body:
; CHECK-UNORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-UNORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP56:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-UNORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-UNORDERED-NEXT: [[TMP57:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-UNORDERED-NEXT: [[MULADD]] = tail call nnan float @llvm.fmuladd.f32(float [[TMP56]], float [[TMP57]], float [[SUM_07]])
; CHECK-UNORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-UNORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-UNORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK-UNORDERED: for.end:
; CHECK-UNORDERED-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ], [ [[TMP55]], [[MIDDLE_BLOCK]] ]
; CHECK-UNORDERED-NEXT: ret float [[MULADD_LCSSA]]
;
; CHECK-ORDERED-LABEL: define float @fmuladd_strict_fmf
; CHECK-ORDERED-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-NEXT: entry:
; CHECK-ORDERED-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-ORDERED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-ORDERED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED: vector.ph:
; CHECK-ORDERED-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 32
; CHECK-ORDERED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-ORDERED-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-ORDERED-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP5:%.*]] = mul i64 [[TMP4]], 32
; CHECK-ORDERED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED: vector.body:
; CHECK-ORDERED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP57:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i32 0
; CHECK-ORDERED-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP28:%.*]] = mul i64 [[TMP27]], 8
; CHECK-ORDERED-NEXT: [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP28]]
; CHECK-ORDERED-NEXT: [[TMP30:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP31:%.*]] = mul i64 [[TMP30]], 16
; CHECK-ORDERED-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP31]]
; CHECK-ORDERED-NEXT: [[TMP33:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP34:%.*]] = mul i64 [[TMP33]], 24
; CHECK-ORDERED-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[TMP22]], i64 [[TMP34]]
; CHECK-ORDERED-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x float>, ptr [[TMP26]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 8 x float>, ptr [[TMP29]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 8 x float>, ptr [[TMP32]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD3:%.*]] = load <vscale x 8 x float>, ptr [[TMP35]], align 4
; CHECK-ORDERED-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP6]]
; CHECK-ORDERED-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i32 0
; CHECK-ORDERED-NEXT: [[TMP41:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP42:%.*]] = mul i64 [[TMP41]], 8
; CHECK-ORDERED-NEXT: [[TMP43:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP42]]
; CHECK-ORDERED-NEXT: [[TMP44:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP45:%.*]] = mul i64 [[TMP44]], 16
; CHECK-ORDERED-NEXT: [[TMP46:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP45]]
; CHECK-ORDERED-NEXT: [[TMP47:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-NEXT: [[TMP48:%.*]] = mul i64 [[TMP47]], 24
; CHECK-ORDERED-NEXT: [[TMP49:%.*]] = getelementptr inbounds float, ptr [[TMP36]], i64 [[TMP48]]
; CHECK-ORDERED-NEXT: [[WIDE_LOAD4:%.*]] = load <vscale x 8 x float>, ptr [[TMP40]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD5:%.*]] = load <vscale x 8 x float>, ptr [[TMP43]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 8 x float>, ptr [[TMP46]], align 4
; CHECK-ORDERED-NEXT: [[WIDE_LOAD7:%.*]] = load <vscale x 8 x float>, ptr [[TMP49]], align 4
; CHECK-ORDERED-NEXT: [[TMP50:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_LOAD]], [[WIDE_LOAD4]]
; CHECK-ORDERED-NEXT: [[TMP51:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_LOAD1]], [[WIDE_LOAD5]]
; CHECK-ORDERED-NEXT: [[TMP52:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_LOAD2]], [[WIDE_LOAD6]]
; CHECK-ORDERED-NEXT: [[TMP53:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_LOAD3]], [[WIDE_LOAD7]]
; CHECK-ORDERED-NEXT: [[TMP54:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[TMP50]])
; CHECK-ORDERED-NEXT: [[TMP55:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP54]], <vscale x 8 x float> [[TMP51]])
; CHECK-ORDERED-NEXT: [[TMP56:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP55]], <vscale x 8 x float> [[TMP52]])
; CHECK-ORDERED-NEXT: [[TMP57]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP56]], <vscale x 8 x float> [[TMP53]])
; CHECK-ORDERED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-ORDERED-NEXT: [[TMP58:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[TMP58]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; CHECK-ORDERED: middle.block:
; CHECK-ORDERED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-ORDERED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED: scalar.ph:
; CHECK-ORDERED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP57]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED: for.body:
; CHECK-ORDERED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP59:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-NEXT: [[TMP60:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-NEXT: [[MULADD]] = tail call nnan float @llvm.fmuladd.f32(float [[TMP59]], float [[TMP60]], float [[SUM_07]])
; CHECK-ORDERED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP19:![0-9]+]]
; CHECK-ORDERED: for.end:
; CHECK-ORDERED-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ], [ [[TMP57]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-NEXT: ret float [[MULADD_LCSSA]]
;
; CHECK-ORDERED-TF-LABEL: define float @fmuladd_strict_fmf
; CHECK-ORDERED-TF-SAME: (ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-ORDERED-TF-NEXT: entry:
; CHECK-ORDERED-TF-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-ORDERED-TF: vector.ph:
; CHECK-ORDERED-TF-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP2:%.*]] = sub i64 [[TMP1]], 1
; CHECK-ORDERED-TF-NEXT: [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP2]]
; CHECK-ORDERED-TF-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP1]]
; CHECK-ORDERED-TF-NEXT: [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
; CHECK-ORDERED-TF-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP4:%.*]] = mul i64 [[TMP3]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 32
; CHECK-ORDERED-TF-NEXT: [[TMP7:%.*]] = sub i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP8:%.*]] = icmp ugt i64 [[N]], [[TMP6]]
; CHECK-ORDERED-TF-NEXT: [[TMP9:%.*]] = select i1 [[TMP8]], i64 [[TMP7]], i64 0
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = mul i64 [[TMP10]], 8
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT:%.*]] = add i64 0, [[TMP11]]
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP13:%.*]] = mul i64 [[TMP12]], 16
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT1:%.*]] = add i64 0, [[TMP13]]
; CHECK-ORDERED-TF-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = mul i64 [[TMP14]], 24
; CHECK-ORDERED-TF-NEXT: [[INDEX_PART_NEXT2:%.*]] = add i64 0, [[TMP15]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 0, i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY3:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY4:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT1]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_ENTRY5:%.*]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_PART_NEXT2]], i64 [[N]])
; CHECK-ORDERED-TF-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-ORDERED-TF: vector.body:
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK6:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY3]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT16:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK7:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY4]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT17:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK8:%.*]] = phi <vscale x 8 x i1> [ [[ACTIVE_LANE_MASK_ENTRY5]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT18:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 0.000000e+00, [[VECTOR_PH]] ], [ [[TMP71:%.*]], [[VECTOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = add i64 [[INDEX]], 0
; CHECK-ORDERED-TF-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP16]]
; CHECK-ORDERED-TF-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i32 0
; CHECK-ORDERED-TF-NEXT: [[TMP37:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP38:%.*]] = mul i64 [[TMP37]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP39:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP38]]
; CHECK-ORDERED-TF-NEXT: [[TMP40:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP41:%.*]] = mul i64 [[TMP40]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP42:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP41]]
; CHECK-ORDERED-TF-NEXT: [[TMP43:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP44:%.*]] = mul i64 [[TMP43]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP45:%.*]] = getelementptr inbounds float, ptr [[TMP32]], i64 [[TMP44]]
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP36]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD9:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP39]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD10:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP42]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD11:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP45]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP46:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP16]]
; CHECK-ORDERED-TF-NEXT: [[TMP50:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i32 0
; CHECK-ORDERED-TF-NEXT: [[TMP51:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP52:%.*]] = mul i64 [[TMP51]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP53:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i64 [[TMP52]]
; CHECK-ORDERED-TF-NEXT: [[TMP54:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP55:%.*]] = mul i64 [[TMP54]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP56:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i64 [[TMP55]]
; CHECK-ORDERED-TF-NEXT: [[TMP57:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP58:%.*]] = mul i64 [[TMP57]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP59:%.*]] = getelementptr inbounds float, ptr [[TMP46]], i64 [[TMP58]]
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD12:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP50]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD13:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP53]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD14:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP56]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD15:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP59]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> poison)
; CHECK-ORDERED-TF-NEXT: [[TMP60:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_MASKED_LOAD]], [[WIDE_MASKED_LOAD12]]
; CHECK-ORDERED-TF-NEXT: [[TMP61:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_MASKED_LOAD9]], [[WIDE_MASKED_LOAD13]]
; CHECK-ORDERED-TF-NEXT: [[TMP62:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_MASKED_LOAD10]], [[WIDE_MASKED_LOAD14]]
; CHECK-ORDERED-TF-NEXT: [[TMP63:%.*]] = fmul nnan <vscale x 8 x float> [[WIDE_MASKED_LOAD11]], [[WIDE_MASKED_LOAD15]]
; CHECK-ORDERED-TF-NEXT: [[TMP64:%.*]] = select nnan <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> [[TMP60]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP65:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[TMP64]])
; CHECK-ORDERED-TF-NEXT: [[TMP66:%.*]] = select nnan <vscale x 8 x i1> [[ACTIVE_LANE_MASK6]], <vscale x 8 x float> [[TMP61]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP67:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP65]], <vscale x 8 x float> [[TMP66]])
; CHECK-ORDERED-TF-NEXT: [[TMP68:%.*]] = select nnan <vscale x 8 x i1> [[ACTIVE_LANE_MASK7]], <vscale x 8 x float> [[TMP62]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP69:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP67]], <vscale x 8 x float> [[TMP68]])
; CHECK-ORDERED-TF-NEXT: [[TMP70:%.*]] = select nnan <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> [[TMP63]], <vscale x 8 x float> shufflevector (<vscale x 8 x float> insertelement (<vscale x 8 x float> poison, float -0.000000e+00, i64 0), <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP71]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP69]], <vscale x 8 x float> [[TMP70]])
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
; CHECK-ORDERED-TF-NEXT: [[TMP72:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP73:%.*]] = mul i64 [[TMP72]], 8
; CHECK-ORDERED-TF-NEXT: [[TMP74:%.*]] = add i64 [[INDEX]], [[TMP73]]
; CHECK-ORDERED-TF-NEXT: [[TMP75:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP76:%.*]] = mul i64 [[TMP75]], 16
; CHECK-ORDERED-TF-NEXT: [[TMP77:%.*]] = add i64 [[INDEX]], [[TMP76]]
; CHECK-ORDERED-TF-NEXT: [[TMP78:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-ORDERED-TF-NEXT: [[TMP79:%.*]] = mul i64 [[TMP78]], 24
; CHECK-ORDERED-TF-NEXT: [[TMP80:%.*]] = add i64 [[INDEX]], [[TMP79]]
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT16]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP74]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT17]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP77]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT18]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP80]], i64 [[TMP9]])
; CHECK-ORDERED-TF-NEXT: [[TMP81:%.*]] = xor <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 8 x i1> insertelement (<vscale x 8 x i1> poison, i1 true, i64 0), <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-ORDERED-TF-NEXT: [[TMP82:%.*]] = extractelement <vscale x 8 x i1> [[TMP81]], i32 0
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP82]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; CHECK-ORDERED-TF: middle.block:
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK-ORDERED-TF: scalar.ph:
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP71]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
; CHECK-ORDERED-TF: for.body:
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[SUM_07:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[MULADD:%.*]], [[FOR_BODY]] ]
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP83:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
; CHECK-ORDERED-TF-NEXT: [[TMP84:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-ORDERED-TF-NEXT: [[MULADD]] = tail call nnan float @llvm.fmuladd.f32(float [[TMP83]], float [[TMP84]], float [[SUM_07]])
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP19:![0-9]+]]
; CHECK-ORDERED-TF: for.end:
; CHECK-ORDERED-TF-NEXT: [[MULADD_LCSSA:%.*]] = phi float [ [[MULADD]], [[FOR_BODY]] ], [ [[TMP71]], [[MIDDLE_BLOCK]] ]
; CHECK-ORDERED-TF-NEXT: ret float [[MULADD_LCSSA]]
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%sum.07 = phi float [ 0.000000e+00, %entry ], [ %muladd, %for.body ]
%arrayidx = getelementptr inbounds float, ptr %a, i64 %iv
%0 = load float, ptr %arrayidx, align 4
%arrayidx2 = getelementptr inbounds float, ptr %b, i64 %iv
%1 = load float, ptr %arrayidx2, align 4
%muladd = tail call nnan float @llvm.fmuladd.f32(float %0, float %1, float %sum.07)
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !1
for.end:
ret float %muladd
}
declare float @llvm.fmuladd.f32(float, float, float)
attributes #0 = { vscale_range(1, 16) }
!0 = distinct !{!0, !3, !6, !8}
!1 = distinct !{!1, !3, !7, !8}
!2 = distinct !{!2, !4, !6, !8}
!3 = !{!"llvm.loop.vectorize.width", i32 8}
!4 = !{!"llvm.loop.vectorize.width", i32 4}
!5 = !{!"llvm.loop.vectorize.width", i32 2}
!6 = !{!"llvm.loop.interleave.count", i32 1}
!7 = !{!"llvm.loop.interleave.count", i32 4}
!8 = !{!"llvm.loop.vectorize.scalable.enable", i1 true}
Codebase Browser
llvm