test/Transforms/LoopVectorize/select-unknown-prof.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --filter "br i1|= select |!prof|!llvm.loop|^.*:" --version 6
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S %s | FileCheck %s

; The select computing the final value of an any-of reduction.
define i32 @anyof_reduction(ptr %a, i64 %n) !prof !0 {
; CHECK-LABEL: define i32 @anyof_reduction(
; CHECK-SAME: ptr [[A:%.*]], i64 [[N:%.*]]) !prof [[PROF0:![0-9]+]] {
; CHECK:  [[ENTRY:.*:]]
; CHECK:    br i1 [[MIN_ITERS_CHECK:%.*]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:  [[VECTOR_PH]]:
; CHECK:  [[VECTOR_BODY:.*]]:
; CHECK:    br i1 [[TMP4:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP1:![0-9]+]]
; CHECK:  [[MIDDLE_BLOCK]]:
; CHECK:    [[RDX_SELECT:%.*]] = select i1 [[TMP6:%.*]], i32 1, i32 0
; CHECK:    br i1 [[CMP_N:%.*]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK:  [[SCALAR_PH]]:
; CHECK:  [[LOOP:.*]]:
; CHECK:    [[RES_NEXT:%.*]] = select i1 [[C:%.*]], i32 1, i32 [[RES:%.*]]
; CHECK:    br i1 [[EC:%.*]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK:  [[EXIT]]:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %res = phi i32 [ 0, %entry ], [ %res.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %a, i64 %iv
  %l = load i32, ptr %gep, align 4
  %c = icmp eq i32 %l, 42
  %res.next = select i1 %c, i32 1, i32 %res
  %iv.next = add nuw nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %res.next
}

; The select computing the final value of a find-last-IV reduction.
define i64 @find_last_iv_reduction(ptr %a, i64 %n) !prof !0 {
; CHECK-LABEL: define i64 @find_last_iv_reduction(
; CHECK-SAME: ptr [[A:%.*]], i64 [[N:%.*]]) !prof [[PROF0]] {
; CHECK:  [[ENTRY:.*:]]
; CHECK:    br i1 [[MIN_ITERS_CHECK:%.*]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:  [[VECTOR_PH]]:
; CHECK:  [[VECTOR_BODY:.*]]:
; CHECK:    [[TMP3:%.*]] = select <4 x i1> [[TMP2:%.*]], <4 x i64> [[VEC_IND:%.*]], <4 x i64> [[VEC_PHI:%.*]]
; CHECK:    br i1 [[TMP4:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK:  [[MIDDLE_BLOCK]]:
; CHECK:    [[TMP7:%.*]] = select i1 [[TMP6:%.*]], i64 [[TMP5:%.*]], i64 -1
; CHECK:    br i1 [[CMP_N:%.*]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK:  [[SCALAR_PH]]:
; CHECK:  [[LOOP:.*]]:
; CHECK:    [[RES_NEXT:%.*]] = select i1 [[C:%.*]], i64 [[IV:%.*]], i64 [[RES:%.*]]
; CHECK:    br i1 [[EC:%.*]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK:  [[EXIT]]:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %res = phi i64 [ -1, %entry ], [ %res.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %a, i64 %iv
  %l = load i32, ptr %gep, align 4
  %c = icmp eq i32 %l, 42
  %res.next = select i1 %c, i64 %iv, i64 %res
  %iv.next = add nuw nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret i64 %res.next
}

; The select rounding the vector trip count down for a scalar epilogue.
define void @requires_scalar_epilogue(ptr noalias %a, ptr noalias %b, i64 %n) !prof !0 {
; CHECK-LABEL: define void @requires_scalar_epilogue(
; CHECK-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]], i64 [[N:%.*]]) !prof [[PROF0]] {
; CHECK:  [[ENTRY:.*:]]
; CHECK:    br i1 [[MIN_ITERS_CHECK:%.*]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:  [[VECTOR_PH]]:
; CHECK:    [[TMP4:%.*]] = select i1 [[TMP3:%.*]], i64 4, i64 [[TMP2:%.*]]
; CHECK:  [[VECTOR_BODY:.*]]:
; CHECK:    br i1 [[TMP7:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK:  [[MIDDLE_BLOCK]]:
; CHECK:  [[SCALAR_PH]]:
; CHECK:  [[LOOP_HEADER:.*]]:
; CHECK:    br i1 [[EC_0:%.*]], label %[[EXIT:.*]], label %[[LOOP_LATCH:.*]]
; CHECK:  [[LOOP_LATCH]]:
; CHECK:    br i1 [[EC_1:%.*]], label %[[LOOP_HEADER]], label %[[EXIT]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK:  [[EXIT]]:
;
entry:
  br label %loop.header

loop.header:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %ec.0 = icmp eq i64 %iv, %n
  br i1 %ec.0, label %exit, label %loop.latch

loop.latch:
  %gep.a = getelementptr inbounds i32, ptr %a, i64 %iv
  %l = load i32, ptr %gep.a, align 4
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  store i32 %l, ptr %gep.b, align 4
  %iv.next = add nuw i64 %iv, 1
  %ec.1 = icmp ult i64 %iv.next, 1000
  br i1 %ec.1, label %loop.header, label %exit

exit:
  ret void
}

; Same, with weights on the latch providing an estimated trip count.
define void @requires_scalar_epilogue_estimated_tc(ptr noalias %a, ptr noalias %b, i64 %n) !prof !0 {
; CHECK-LABEL: define void @requires_scalar_epilogue_estimated_tc(
; CHECK-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]], i64 [[N:%.*]]) !prof [[PROF0]] {
; CHECK:  [[ENTRY:.*:]]
; CHECK:    br i1 [[MIN_ITERS_CHECK:%.*]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]], !prof [[PROF9:![0-9]+]]
; CHECK:  [[VECTOR_PH]]:
; CHECK:    [[TMP4:%.*]] = select i1 [[TMP3:%.*]], i64 4, i64 [[TMP2:%.*]]
; CHECK:  [[VECTOR_BODY:.*]]:
; CHECK:    br i1 [[TMP7:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !prof [[PROF10:![0-9]+]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK:  [[MIDDLE_BLOCK]]:
; CHECK:  [[SCALAR_PH]]:
; CHECK:  [[LOOP_HEADER:.*]]:
; CHECK:    br i1 [[EC_0:%.*]], label %[[EXIT:.*]], label %[[LOOP_LATCH:.*]]
; CHECK:  [[LOOP_LATCH]]:
; CHECK:    br i1 [[EC_1:%.*]], label %[[LOOP_HEADER]], label %[[EXIT]], !prof [[PROF13:![0-9]+]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK:  [[EXIT]]:
;
entry:
  br label %loop.header

loop.header:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %ec.0 = icmp eq i64 %iv, %n
  br i1 %ec.0, label %exit, label %loop.latch

loop.latch:
  %gep.a = getelementptr inbounds i32, ptr %a, i64 %iv
  %l = load i32, ptr %gep.a, align 4
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  store i32 %l, ptr %gep.b, align 4
  %iv.next = add nuw i64 %iv, 1
  %ec.1 = icmp ult i64 %iv.next, 1000
  br i1 %ec.1, label %loop.header, label %exit, !prof !1

exit:
  ret void
}

define i32 @anyof_reduction_not_profiled(ptr %a, i64 %n) {
; CHECK-LABEL: define i32 @anyof_reduction_not_profiled(
; CHECK-SAME: ptr [[A:%.*]], i64 [[N:%.*]]) {
; CHECK:  [[ENTRY:.*:]]
; CHECK:    br i1 [[MIN_ITERS_CHECK:%.*]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:  [[VECTOR_PH]]:
; CHECK:  [[VECTOR_BODY:.*]]:
; CHECK:    br i1 [[TMP4:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK:  [[MIDDLE_BLOCK]]:
; CHECK:    [[RDX_SELECT:%.*]] = select i1 [[TMP6:%.*]], i32 1, i32 0
; CHECK:    br i1 [[CMP_N:%.*]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK:  [[SCALAR_PH]]:
; CHECK:  [[LOOP:.*]]:
; CHECK:    [[RES_NEXT:%.*]] = select i1 [[C:%.*]], i32 1, i32 [[RES:%.*]]
; CHECK:    br i1 [[EC:%.*]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK:  [[EXIT]]:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %res = phi i32 [ 0, %entry ], [ %res.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %a, i64 %iv
  %l = load i32, ptr %gep, align 4
  %c = icmp eq i32 %l, 42
  %res.next = select i1 %c, i32 1, i32 %res
  %iv.next = add nuw nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %res.next
}

!0 = !{!"function_entry_count", i64 1000}
!1 = !{!"branch_weights", i32 999, i32 1}
;.
; CHECK: [[PROF0]] = !{!"function_entry_count", i64 1000}
; CHECK: [[LOOP1]] = distinct !{[[LOOP1]], [[META2:![0-9]+]], [[META3:![0-9]+]]}
; CHECK: [[META2]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META3]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META3]], [[META2]]}
; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META3]]}
; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META3]], [[META2]]}
; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META3]]}
; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META3]], [[META2]]}
; CHECK: [[PROF9]] = !{!"branch_weights", i32 1, i32 127}
; CHECK: [[PROF10]] = !{!"branch_weights", i32 1, i32 249}
; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META2]], [[META3]], [[META12:![0-9]+]]}
; CHECK: [[META12]] = !{!"llvm.loop.estimated_trip_count", i32 250}
; CHECK: [[PROF13]] = !{!"branch_weights", i32 0, i32 1}
; CHECK: [[LOOP14]] = distinct !{[[LOOP14]], [[META3]], [[META2]], [[META15:![0-9]+]]}
; CHECK: [[META15]] = !{!"llvm.loop.estimated_trip_count", i32 0}
; CHECK: [[LOOP16]] = distinct !{[[LOOP16]], [[META2]], [[META3]]}
; CHECK: [[LOOP17]] = distinct !{[[LOOP17]], [[META3]], [[META2]]}
;.