llvm/test/Transforms/LoopVectorize/early-exit-unary-ops.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -S %s | FileCheck %s

define i64 @early_exit_with_fneg(ptr dereferenceable(1024) align 8 %src, i1 %cond) {
; CHECK-LABEL: define i64 @early_exit_with_fneg(
; CHECK-SAME: ptr align 8 dereferenceable(1024) [[SRC:%.*]], i1 [[COND:%.*]]) {
; CHECK-NEXT:  [[LOOP:.*:]]
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    br label %[[VECTOR_BODY1:.*]]
; CHECK:       [[VECTOR_BODY1]]:
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, %[[VECTOR_BODY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY_INTERIM:.*]] ]
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds double, ptr [[SRC]], i64 [[IV]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[GEP]], align 8
; CHECK-NEXT:    [[TMP1:%.*]] = fneg <4 x double> [[WIDE_LOAD]]
; CHECK-NEXT:    [[TMP2:%.*]] = fcmp oeq <4 x double> [[TMP1]], splat (double 1.000000e+01)
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[IV]], 4
; CHECK-NEXT:    [[TMP3:%.*]] = freeze <4 x i1> [[TMP2]]
; CHECK-NEXT:    [[TMP4:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP3]])
; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 124
; CHECK-NEXT:    br i1 [[TMP4]], label %[[VECTOR_EARLY_EXIT:.*]], label %[[VECTOR_BODY_INTERIM]]
; CHECK:       [[VECTOR_BODY_INTERIM]]:
; CHECK-NEXT:    br i1 [[TMP5]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY1]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[SCALAR_PH:.*]]
; CHECK:       [[VECTOR_EARLY_EXIT]]:
; CHECK-NEXT:    [[TMP6:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP2]], i1 false)
; CHECK-NEXT:    [[TMP7:%.*]] = add i64 [[IV]], [[TMP6]]
; CHECK-NEXT:    br label %[[EARLY_EXIT:.*]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    br label %[[LOOP_HEADER:.*]]
; CHECK:       [[LOOP_HEADER]]:
; CHECK-NEXT:    [[IV1:%.*]] = phi i64 [ [[IV_NEXT:%.*]], %[[LATCH:.*]] ], [ 124, %[[SCALAR_PH]] ]
; CHECK-NEXT:    [[GEP1:%.*]] = getelementptr inbounds double, ptr [[SRC]], i64 [[IV1]]
; CHECK-NEXT:    [[VAL:%.*]] = load double, ptr [[GEP1]], align 8
; CHECK-NEXT:    [[NEG:%.*]] = fneg double [[VAL]]
; CHECK-NEXT:    [[C_1:%.*]] = fcmp une double [[NEG]], 1.000000e+01
; CHECK-NEXT:    br i1 [[C_1]], label %[[LATCH]], label %[[EARLY_EXIT]]
; CHECK:       [[LATCH]]:
; CHECK-NEXT:    [[IV_NEXT]] = add nuw i64 [[IV1]], 1
; CHECK-NEXT:    [[EXIT_COND:%.*]] = icmp eq i64 [[IV_NEXT]], 127
; CHECK-NEXT:    br i1 [[EXIT_COND]], label %[[EXIT:.*]], label %[[LOOP_HEADER]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       [[EARLY_EXIT]]:
; CHECK-NEXT:    [[IV_LCSSA:%.*]] = phi i64 [ [[IV1]], %[[LOOP_HEADER]] ], [ [[TMP7]], %[[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT:    ret i64 [[IV_LCSSA]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret i64 10
;
entry:
  br label %loop.header

loop.header:
  %iv = phi i64 [ %iv.next, %latch ], [ 0, %entry ]
  %gep = getelementptr inbounds double, ptr %src, i64 %iv
  %val = load double, ptr %gep, align 8
  %neg = fneg double %val
  %c.1 = fcmp une double %neg, 10.0
  br i1 %c.1, label %latch, label %early.exit

latch:
  %iv.next = add nuw i64 %iv, 1
  %exit.cond = icmp eq i64 %iv.next, 127
  br i1 %exit.cond, label %exit, label %loop.header

early.exit:
  ret i64 %iv

exit:
  ret i64 10
}