test/Transforms/LoopVectorize/find-last.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S %s | FileCheck %s
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -prefer-inloop-reductions -S %s | FileCheck %s

; Test vectorization of a "find last" pattern where a scalar value is
; conditionally updated based on comparing elements from two arrays.
define i32 @find_last_with_select(ptr noalias %a, ptr noalias %b) {
; CHECK-LABEL: @find_last_with_select(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1), [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT:    [[TMP3:%.*]] = icmp sge <4 x i32> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-NEXT:    [[TMP4:%.*]] = freeze <4 x i1> [[TMP3]]
; CHECK-NEXT:    [[TMP5:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP4]])
; CHECK-NEXT:    [[TMP6]] = select i1 [[TMP5]], <4 x i1> [[TMP3]], <4 x i1> [[TMP0]]
; CHECK-NEXT:    [[TMP7]] = select i1 [[TMP5]], <4 x i32> [[WIDE_LOAD]], <4 x i32> [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
; CHECK-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    [[TMP9:%.*]] = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> [[TMP7]], <4 x i1> [[TMP6]], i32 1)
; CHECK-NEXT:    br label [[EXIT:%.*]]
; CHECK:       exit:
; CHECK-NEXT:    ret i32 [[TMP9]]
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %rdx = phi i32 [ 1, %entry ], [ %sel, %loop ]
  %gep.a = getelementptr inbounds i32, ptr %a, i64 %iv
  %load.a = load i32, ptr %gep.a, align 4
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  %load.b = load i32, ptr %gep.b, align 4
  %cmp = icmp slt i32 %load.a, %load.b
  %sel = select i1 %cmp, i32 %rdx, i32 %load.a
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, 500
  br i1 %exitcond, label %exit, label %loop

exit:
  ret i32 %sel
}

; Keep when a[i] > b[i] (signed), i.e., update when a[i] <= b[i].
define i32 @find_last_sgt_inverted(ptr noalias %a, ptr noalias %b) {
; CHECK-LABEL: @find_last_sgt_inverted(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1), [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT:    [[TMP3:%.*]] = icmp sle <4 x i32> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-NEXT:    [[TMP4:%.*]] = freeze <4 x i1> [[TMP3]]
; CHECK-NEXT:    [[TMP5:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP4]])
; CHECK-NEXT:    [[TMP6]] = select i1 [[TMP5]], <4 x i1> [[TMP3]], <4 x i1> [[TMP0]]
; CHECK-NEXT:    [[TMP7]] = select i1 [[TMP5]], <4 x i32> [[WIDE_LOAD]], <4 x i32> [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
; CHECK-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    [[TMP9:%.*]] = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> [[TMP7]], <4 x i1> [[TMP6]], i32 1)
; CHECK-NEXT:    br label [[EXIT:%.*]]
; CHECK:       exit:
; CHECK-NEXT:    ret i32 [[TMP9]]
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %rdx = phi i32 [ 1, %entry ], [ %sel, %loop ]
  %gep.a = getelementptr inbounds i32, ptr %a, i64 %iv
  %load.a = load i32, ptr %gep.a, align 4
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  %load.b = load i32, ptr %gep.b, align 4
  %cmp = icmp sgt i32 %load.a, %load.b
  %sel = select i1 %cmp, i32 %rdx, i32 %load.a
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, 500
  br i1 %exitcond, label %exit, label %loop

exit:
  ret i32 %sel
}

; Keep when a[i] < b[i] (unsigned), i.e., update when a[i] >= b[i].
define i32 @find_last_ult_inverted(ptr noalias %a, ptr noalias %b) {
; CHECK-LABEL: @find_last_ult_inverted(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1), [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT:    [[TMP3:%.*]] = icmp uge <4 x i32> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-NEXT:    [[TMP4:%.*]] = freeze <4 x i1> [[TMP3]]
; CHECK-NEXT:    [[TMP5:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP4]])
; CHECK-NEXT:    [[TMP6]] = select i1 [[TMP5]], <4 x i1> [[TMP3]], <4 x i1> [[TMP0]]
; CHECK-NEXT:    [[TMP7]] = select i1 [[TMP5]], <4 x i32> [[WIDE_LOAD]], <4 x i32> [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
; CHECK-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    [[TMP9:%.*]] = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> [[TMP7]], <4 x i1> [[TMP6]], i32 1)
; CHECK-NEXT:    br label [[EXIT:%.*]]
; CHECK:       exit:
; CHECK-NEXT:    ret i32 [[TMP9]]
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %rdx = phi i32 [ 1, %entry ], [ %sel, %loop ]
  %gep.a = getelementptr inbounds i32, ptr %a, i64 %iv
  %load.a = load i32, ptr %gep.a, align 4
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  %load.b = load i32, ptr %gep.b, align 4
  %cmp = icmp ult i32 %load.a, %load.b
  %sel = select i1 %cmp, i32 %rdx, i32 %load.a
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, 500
  br i1 %exitcond, label %exit, label %loop

exit:
  ret i32 %sel
}

; Keep when a[i] != b[i], i.e., update when a[i] == b[i].
define i32 @find_last_ne_inverted(ptr noalias %a, ptr noalias %b) {
; CHECK-LABEL: @find_last_ne_inverted(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1), [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-NEXT:    [[TMP4:%.*]] = freeze <4 x i1> [[TMP3]]
; CHECK-NEXT:    [[TMP5:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP4]])
; CHECK-NEXT:    [[TMP6]] = select i1 [[TMP5]], <4 x i1> [[TMP3]], <4 x i1> [[TMP0]]
; CHECK-NEXT:    [[TMP7]] = select i1 [[TMP5]], <4 x i32> [[WIDE_LOAD]], <4 x i32> [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
; CHECK-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    [[TMP9:%.*]] = call i32 @llvm.experimental.vector.extract.last.active.v4i32(<4 x i32> [[TMP7]], <4 x i1> [[TMP6]], i32 1)
; CHECK-NEXT:    br label [[EXIT:%.*]]
; CHECK:       exit:
; CHECK-NEXT:    ret i32 [[TMP9]]
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %rdx = phi i32 [ 1, %entry ], [ %sel, %loop ]
  %gep.a = getelementptr inbounds i32, ptr %a, i64 %iv
  %load.a = load i32, ptr %gep.a, align 4
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  %load.b = load i32, ptr %gep.b, align 4
  %cmp = icmp ne i32 %load.a, %load.b
  %sel = select i1 %cmp, i32 %rdx, i32 %load.a
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, 500
  br i1 %exitcond, label %exit, label %loop

exit:
  ret i32 %sel
}