llvm/test/Transforms/LoopVectorize/check-prof-info.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --filter "br" --filter "^.*:" --version 6
; RUN: opt -passes="print<block-freq>,loop-vectorize" -force-vector-width=4 -force-vector-interleave=1 -S < %s |  FileCheck %s
; RUN: opt -passes="print<block-freq>,loop-vectorize" -force-vector-width=4 -force-vector-interleave=4 -S < %s |  FileCheck %s -check-prefix=CHECK-MASKED
; RUN: opt -passes="print<block-freq>,loop-vectorize" -force-vector-width=4 -force-vector-interleave=1 \
; RUN:   -scalable-vectorization=on -force-target-supports-scalable-vectors -S < %s |  FileCheck %s -check-prefix=CHECK-SCALABLE

target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"

@a = global [1024 x i32] zeroinitializer, align 16
@b = global [1024 x i32] zeroinitializer, align 16

; Check correctness of profile info for vectorization without epilog.
define void @_Z3foov() {
; CHECK-LABEL: define void @_Z3foov() {
; CHECK:  [[ENTRY:.*:]]
; CHECK:    br label %[[VECTOR_PH:.*]]
; CHECK:  [[VECTOR_PH]]:
; CHECK:    br label %[[VECTOR_BODY:.*]]
; CHECK:  [[VECTOR_BODY]]:
; CHECK:    br i1 [[TMP4:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !prof [[PROF0:![0-9]+]], !llvm.loop [[LOOP1:![0-9]+]]
; CHECK:  [[MIDDLE_BLOCK]]:
; CHECK:    br label %[[FOR_COND_CLEANUP:.*]]
; CHECK:  [[FOR_COND_CLEANUP]]:
;
; CHECK-MASKED-LABEL: define void @_Z3foov() {
; CHECK-MASKED:  [[ENTRY:.*:]]
; CHECK-MASKED:    br label %[[VECTOR_PH:.*]]
; CHECK-MASKED:  [[VECTOR_PH]]:
; CHECK-MASKED:    br label %[[VECTOR_BODY:.*]]
; CHECK-MASKED:  [[VECTOR_BODY]]:
; CHECK-MASKED:    br i1 [[TMP16:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !prof [[PROF0:![0-9]+]], !llvm.loop [[LOOP1:![0-9]+]]
; CHECK-MASKED:  [[MIDDLE_BLOCK]]:
; CHECK-MASKED:    br label %[[FOR_COND_CLEANUP:.*]]
; CHECK-MASKED:  [[FOR_COND_CLEANUP]]:
;
; CHECK-SCALABLE-LABEL: define void @_Z3foov() {
; CHECK-SCALABLE:  [[ENTRY:.*:]]
; CHECK-SCALABLE:    br i1 [[MIN_ITERS_CHECK:%.*]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]], !prof [[PROF0:![0-9]+]]
; CHECK-SCALABLE:  [[VECTOR_PH]]:
; CHECK-SCALABLE:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[TMP9:%.*]], i64 0
; CHECK-SCALABLE:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-SCALABLE:    br label %[[VECTOR_BODY:.*]]
; CHECK-SCALABLE:  [[VECTOR_BODY]]:
; CHECK-SCALABLE:    [[VEC_IND_NEXT:%.*]] = add <vscale x 4 x i32> [[VEC_IND:%.*]], [[BROADCAST_SPLAT]]
; CHECK-SCALABLE:    br i1 [[TMP10:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !prof [[PROF1:![0-9]+]], !llvm.loop [[LOOP2:![0-9]+]]
; CHECK-SCALABLE:  [[MIDDLE_BLOCK]]:
; CHECK-SCALABLE:    br i1 [[CMP_N:%.*]], label %[[FOR_COND_CLEANUP:.*]], label %[[SCALAR_PH]], !prof [[PROF6:![0-9]+]]
; CHECK-SCALABLE:  [[SCALAR_PH]]:
; CHECK-SCALABLE:    br label %[[FOR_BODY:.*]]
; CHECK-SCALABLE:  [[FOR_COND_CLEANUP]]:
; CHECK-SCALABLE:  [[FOR_BODY]]:
; CHECK-SCALABLE:    br i1 [[EXITCOND:%.*]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]], !prof [[PROF7:![0-9]+]], !llvm.loop [[LOOP8:![0-9]+]]
;
entry:
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds [1024 x i32], ptr @b, i64 0, i64 %iv
  %0 = load i32, ptr %arrayidx, align 4
  %1 = trunc i64 %iv to i32
  %mul = mul nsw i32 %0, %1
  %arrayidx2 = getelementptr inbounds [1024 x i32], ptr @a, i64 0, i64 %iv
  %2 = load i32, ptr %arrayidx2, align 4
  %add = add nsw i32 %2, %mul
  store i32 %add, ptr %arrayidx2, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, 1024
  br i1 %exitcond, label %for.cond.cleanup, label %for.body, !prof !0
}

; Check correctness of profile info for vectorization with epilog.
define void @_Z3foo2v() {
; CHECK-LABEL: define void @_Z3foo2v() {
; CHECK:  [[ENTRY:.*:]]
; CHECK:    br label %[[VECTOR_PH:.*]]
; CHECK:  [[VECTOR_PH]]:
; CHECK:    br label %[[VECTOR_BODY:.*]]
; CHECK:  [[VECTOR_BODY]]:
; CHECK:    br i1 [[TMP4:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !prof [[PROF0]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK:  [[MIDDLE_BLOCK]]:
; CHECK:    br label %[[SCALAR_PH:.*]]
; CHECK:  [[SCALAR_PH]]:
; CHECK:    br label %[[FOR_BODY:.*]]
; CHECK:  [[FOR_COND_CLEANUP:.*]]:
; CHECK:  [[FOR_BODY]]:
; CHECK:    br i1 [[EXITCOND:%.*]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]], !prof [[PROF6:![0-9]+]], !llvm.loop [[LOOP7:![0-9]+]]
;
; CHECK-MASKED-LABEL: define void @_Z3foo2v() {
; CHECK-MASKED:  [[ENTRY:.*:]]
; CHECK-MASKED:    br label %[[VECTOR_PH:.*]]
; CHECK-MASKED:  [[VECTOR_PH]]:
; CHECK-MASKED:    br label %[[VECTOR_BODY:.*]]
; CHECK-MASKED:  [[VECTOR_BODY]]:
; CHECK-MASKED:    br i1 [[TMP16:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !prof [[PROF0]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK-MASKED:  [[MIDDLE_BLOCK]]:
; CHECK-MASKED:    br label %[[SCALAR_PH:.*]]
; CHECK-MASKED:  [[SCALAR_PH]]:
; CHECK-MASKED:    br label %[[FOR_BODY:.*]]
; CHECK-MASKED:  [[FOR_COND_CLEANUP:.*]]:
; CHECK-MASKED:  [[FOR_BODY]]:
; CHECK-MASKED:    br i1 [[EXITCOND:%.*]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]], !prof [[PROF6:![0-9]+]], !llvm.loop [[LOOP7:![0-9]+]]
;
; CHECK-SCALABLE-LABEL: define void @_Z3foo2v() {
; CHECK-SCALABLE:  [[ENTRY:.*:]]
; CHECK-SCALABLE:    br i1 [[MIN_ITERS_CHECK:%.*]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]], !prof [[PROF0]]
; CHECK-SCALABLE:  [[VECTOR_PH]]:
; CHECK-SCALABLE:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[TMP9:%.*]], i64 0
; CHECK-SCALABLE:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-SCALABLE:    br label %[[VECTOR_BODY:.*]]
; CHECK-SCALABLE:  [[VECTOR_BODY]]:
; CHECK-SCALABLE:    [[VEC_IND_NEXT:%.*]] = add <vscale x 4 x i32> [[VEC_IND:%.*]], [[BROADCAST_SPLAT]]
; CHECK-SCALABLE:    br i1 [[TMP10:%.*]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !prof [[PROF1]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK-SCALABLE:  [[MIDDLE_BLOCK]]:
; CHECK-SCALABLE:    br i1 [[CMP_N:%.*]], label %[[FOR_COND_CLEANUP:.*]], label %[[SCALAR_PH]], !prof [[PROF6]]
; CHECK-SCALABLE:  [[SCALAR_PH]]:
; CHECK-SCALABLE:    br label %[[FOR_BODY:.*]]
; CHECK-SCALABLE:  [[FOR_COND_CLEANUP]]:
; CHECK-SCALABLE:  [[FOR_BODY]]:
; CHECK-SCALABLE:    br i1 [[EXITCOND:%.*]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]], !prof [[PROF11:![0-9]+]], !llvm.loop [[LOOP12:![0-9]+]]
;
entry:
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
  %arrayidx = getelementptr inbounds [1024 x i32], ptr @b, i64 0, i64 %iv
  %0 = load i32, ptr %arrayidx, align 4
  %1 = trunc i64 %iv to i32
  %mul = mul nsw i32 %0, %1
  %arrayidx2 = getelementptr inbounds [1024 x i32], ptr @a, i64 0, i64 %iv
  %2 = load i32, ptr %arrayidx2, align 4
  %add = add nsw i32 %2, %mul
  store i32 %add, ptr %arrayidx2, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, 1027
  br i1 %exitcond, label %for.cond.cleanup, label %for.body, !prof !1
}


!0 = !{!"branch_weights", i32 1, i32 1023}
!1 = !{!"branch_weights", i32 1, i32 1026}
;.
; CHECK: [[PROF0]] = !{!"branch_weights", i32 1, i32 255}
; CHECK: [[LOOP1]] = distinct !{[[LOOP1]], [[META2:![0-9]+]], [[META3:![0-9]+]], [[META4:![0-9]+]]}
; CHECK: [[META2]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META3]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK: [[META4]] = !{!"llvm.loop.estimated_trip_count", i32 256}
; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META3]], [[META4]]}
; CHECK: [[PROF6]] = !{!"branch_weights", i32 1, i32 2}
; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META3]], [[META2]], [[META8:![0-9]+]]}
; CHECK: [[META8]] = !{!"llvm.loop.estimated_trip_count", i32 3}
;.
; CHECK-MASKED: [[PROF0]] = !{!"branch_weights", i32 1, i32 63}
; CHECK-MASKED: [[LOOP1]] = distinct !{[[LOOP1]], [[META2:![0-9]+]], [[META3:![0-9]+]], [[META4:![0-9]+]]}
; CHECK-MASKED: [[META2]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK-MASKED: [[META3]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK-MASKED: [[META4]] = !{!"llvm.loop.estimated_trip_count", i32 64}
; CHECK-MASKED: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META3]], [[META4]]}
; CHECK-MASKED: [[PROF6]] = !{!"branch_weights", i32 1, i32 2}
; CHECK-MASKED: [[LOOP7]] = distinct !{[[LOOP7]], [[META3]], [[META2]], [[META8:![0-9]+]]}
; CHECK-MASKED: [[META8]] = !{!"llvm.loop.estimated_trip_count", i32 3}
;.
; CHECK-SCALABLE: [[PROF0]] = !{!"branch_weights", i32 1, i32 127}
; CHECK-SCALABLE: [[PROF1]] = !{!"branch_weights", i32 1, i32 255}
; CHECK-SCALABLE: [[LOOP2]] = distinct !{[[LOOP2]], [[META3:![0-9]+]], [[META4:![0-9]+]], [[META5:![0-9]+]]}
; CHECK-SCALABLE: [[META3]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK-SCALABLE: [[META4]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK-SCALABLE: [[META5]] = !{!"llvm.loop.estimated_trip_count", i32 256}
; CHECK-SCALABLE: [[PROF6]] = !{!"branch_weights", i32 1, i32 3}
; CHECK-SCALABLE: [[PROF7]] = !{!"branch_weights", i32 1, i32 0}
; CHECK-SCALABLE: [[LOOP8]] = distinct !{[[LOOP8]], [[META4]], [[META3]], [[META9:![0-9]+]]}
; CHECK-SCALABLE: [[META9]] = !{!"llvm.loop.estimated_trip_count", i32 0}
; CHECK-SCALABLE: [[LOOP10]] = distinct !{[[LOOP10]], [[META3]], [[META4]], [[META5]]}
; CHECK-SCALABLE: [[PROF11]] = !{!"branch_weights", i32 1, i32 2}
; CHECK-SCALABLE: [[LOOP12]] = distinct !{[[LOOP12]], [[META4]], [[META3]], [[META13:![0-9]+]]}
; CHECK-SCALABLE: [[META13]] = !{!"llvm.loop.estimated_trip_count", i32 3}
;.