llvm/test/Transforms/LoopVectorize/AArch64/invalid-costs.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "scalar.ph:" --version 6
; RUN: opt -passes="loop-vectorize" -pass-remarks-output=%t.yaml -S %s | FileCheck %s
; RUN: FileCheck --input-file=%t.yaml --check-prefix=REMARKS %s

target triple = "arm64-apple-macosx"

; REMARKS: Recipe with invalid costs prevented vectorization at VF=(vscale x 1): load
; Test case for https://github.com/llvm/llvm-project/issues/160792.
define void @replicate_sdiv_conditional(ptr noalias %a, ptr noalias %b, ptr noalias %c) #0 {
; CHECK-LABEL: define void @replicate_sdiv_conditional(
; CHECK-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]], ptr noalias [[C:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT:    [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 2
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 64, [[TMP1]]
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT:    [[TMP3:%.*]] = shl nuw i64 [[TMP2]], 2
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 64, [[TMP3]]
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 64, [[N_MOD_VF]]
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT:    [[TMP5:%.*]] = icmp slt <vscale x 4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr i32, ptr [[B]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP6]], <vscale x 4 x i1> [[TMP5]], <vscale x 4 x i32> poison)
; CHECK-NEXT:    [[TMP7:%.*]] = sext <vscale x 4 x i32> [[WIDE_MASKED_LOAD]] to <vscale x 4 x i64>
; CHECK-NEXT:    [[TMP8:%.*]] = ashr <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], splat (i32 1)
; CHECK-NEXT:    [[TMP9:%.*]] = add <vscale x 4 x i32> [[TMP8]], [[WIDE_LOAD]]
; CHECK-NEXT:    [[TMP10:%.*]] = sext <vscale x 4 x i32> [[TMP9]] to <vscale x 4 x i64>
; CHECK-NEXT:    [[TMP11:%.*]] = select <vscale x 4 x i1> [[TMP5]], <vscale x 4 x i64> [[TMP7]], <vscale x 4 x i64> splat (i64 1)
; CHECK-NEXT:    [[TMP12:%.*]] = sdiv <vscale x 4 x i64> [[TMP10]], [[TMP11]]
; CHECK-NEXT:    [[TMP13:%.*]] = trunc <vscale x 4 x i64> [[TMP12]] to <vscale x 4 x i32>
; CHECK-NEXT:    [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP5]], <vscale x 4 x i32> [[TMP13]], <vscale x 4 x i32> [[WIDE_LOAD]]
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT:    store <vscale x 4 x i32> [[PREDPHI]], ptr [[TMP14]], align 4
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP15]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 64, [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], [[FOR_END:label %.*]], label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  br label %loop.header

loop.header:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %gep.c = getelementptr inbounds i32, ptr %c, i64 %iv
  %val.c = load i32, ptr %gep.c, align 4
  %cmp = icmp slt i32 %val.c, 0
  br i1 %cmp, label %if.then, label %loop.latch

if.then:
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  %val.b = load i32, ptr %gep.b, align 4
  %sext = sext i32 %val.b to i64
  %shr = ashr i32 %val.b, 1
  %add = add i32 %shr, %val.c
  %conv = sext i32 %add to i64
  %div = sdiv i64 %conv, %sext
  %trunc = trunc i64 %div to i32
  br label %loop.latch

loop.latch:
  %result = phi i32 [ %trunc, %if.then ], [ %val.c, %loop.header ]
  %gep.a = getelementptr inbounds i32, ptr %a, i64 %iv
  store i32 %result, ptr %gep.a, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %exit = icmp eq i64 %iv.next, 64
  br i1 %exit, label %for.end, label %loop.header

for.end:
  ret void
}

attributes #0 = { "target-cpu"="neoverse-512tvb" }