blob: 2d398787aeb5fce7a919a43e48ba9c3d0153f264 [file]
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --filter-out-after "^scalar.ph:" --version 6
; RUN: opt -passes=loop-vectorize -mattr=+sve -tail-folding-policy=dont-fold-tail -S %s | FileCheck %s
target triple = "aarch64-unknown-linux-gnu"
; Test case where the minimum profitable trip count due to runtime checks
; exceeds VF.getKnownMinValue() * UF.
define void @min_trip_count_due_to_runtime_checks_1(ptr %dst.1, ptr %dst.2, ptr %src.1, ptr %src.2, i64 %n) vscale_range(1,16) {
; CHECK-LABEL: define void @min_trip_count_due_to_runtime_checks_1(
; CHECK-SAME: ptr [[DST_1:%.*]], ptr [[DST_2:%.*]], ptr [[SRC_1:%.*]], ptr [[SRC_2:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[SRC_215:%.*]] = ptrtoaddr ptr [[SRC_2]] to i64
; CHECK-NEXT: [[SRC_114:%.*]] = ptrtoaddr ptr [[SRC_1]] to i64
; CHECK-NEXT: [[DST_113:%.*]] = ptrtoaddr ptr [[DST_1]] to i64
; CHECK-NEXT: [[DST_212:%.*]] = ptrtoaddr ptr [[DST_2]] to i64
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.umax.i64(i64 [[N]], i64 1)
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP2:%.*]] = shl nuw i64 [[TMP1]], 2
; CHECK-NEXT: [[TMP3:%.*]] = call i64 @llvm.umax.i64(i64 [[TMP2]], i64 20)
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], [[TMP3]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK: [[VECTOR_MEMCHECK]]:
; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[TMP1]], 5
; CHECK-NEXT: [[TMP5:%.*]] = add nsw i64 [[TMP4]], -1
; CHECK-NEXT: [[TMP6:%.*]] = sub i64 [[DST_212]], [[DST_113]]
; CHECK-NEXT: [[TMP7:%.*]] = sub i64 [[TMP6]], 1
; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP7]], [[TMP5]]
; CHECK-NEXT: [[TMP8:%.*]] = sub i64 [[DST_113]], [[SRC_114]]
; CHECK-NEXT: [[TMP9:%.*]] = sub i64 [[TMP8]], 1
; CHECK-NEXT: [[DIFF_CHECK16:%.*]] = icmp ult i64 [[TMP9]], [[TMP5]]
; CHECK-NEXT: [[CONFLICT_RDX:%.*]] = or i1 [[DIFF_CHECK]], [[DIFF_CHECK16]]
; CHECK-NEXT: [[TMP10:%.*]] = sub i64 [[DST_113]], [[SRC_215]]
; CHECK-NEXT: [[TMP11:%.*]] = sub i64 [[TMP10]], 1
; CHECK-NEXT: [[DIFF_CHECK17:%.*]] = icmp ult i64 [[TMP11]], [[TMP5]]
; CHECK-NEXT: [[CONFLICT_RDX18:%.*]] = or i1 [[CONFLICT_RDX]], [[DIFF_CHECK17]]
; CHECK-NEXT: [[TMP12:%.*]] = sub i64 [[DST_212]], [[SRC_114]]
; CHECK-NEXT: [[TMP13:%.*]] = sub i64 [[TMP12]], 1
; CHECK-NEXT: [[DIFF_CHECK19:%.*]] = icmp ult i64 [[TMP13]], [[TMP5]]
; CHECK-NEXT: [[CONFLICT_RDX20:%.*]] = or i1 [[CONFLICT_RDX18]], [[DIFF_CHECK19]]
; CHECK-NEXT: [[TMP14:%.*]] = sub i64 [[DST_212]], [[SRC_215]]
; CHECK-NEXT: [[TMP15:%.*]] = sub i64 [[TMP14]], 1
; CHECK-NEXT: [[DIFF_CHECK21:%.*]] = icmp ult i64 [[TMP15]], [[TMP5]]
; CHECK-NEXT: [[CONFLICT_RDX22:%.*]] = or i1 [[CONFLICT_RDX20]], [[DIFF_CHECK21]]
; CHECK-NEXT: br i1 [[CONFLICT_RDX22]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: [[TMP16:%.*]] = shl nuw i64 [[TMP1]], 1
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], [[TMP2]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[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: [[TMP17:%.*]] = getelementptr i64, ptr [[SRC_1]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i64, ptr [[SRC_2]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP19:%.*]] = getelementptr [8 x i8], ptr [[TMP17]], i64 [[TMP16]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP17]], align 8
; CHECK-NEXT: [[WIDE_LOAD23:%.*]] = load <vscale x 2 x i64>, ptr [[TMP19]], align 8
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr [8 x i8], ptr [[TMP18]], i64 [[TMP16]]
; CHECK-NEXT: [[WIDE_LOAD24:%.*]] = load <vscale x 2 x i64>, ptr [[TMP18]], align 8
; CHECK-NEXT: [[WIDE_LOAD25:%.*]] = load <vscale x 2 x i64>, ptr [[TMP20]], align 8
; CHECK-NEXT: [[TMP21:%.*]] = add <vscale x 2 x i64> [[WIDE_LOAD]], [[WIDE_LOAD24]]
; CHECK-NEXT: [[TMP22:%.*]] = add <vscale x 2 x i64> [[WIDE_LOAD23]], [[WIDE_LOAD25]]
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i64, ptr [[DST_1]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i64, ptr [[DST_2]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP25:%.*]] = getelementptr [8 x i8], ptr [[TMP23]], i64 [[TMP16]]
; CHECK-NEXT: store <vscale x 2 x i64> [[TMP21]], ptr [[TMP23]], align 8
; CHECK-NEXT: store <vscale x 2 x i64> [[TMP22]], ptr [[TMP25]], align 8
; CHECK-NEXT: [[TMP26:%.*]] = getelementptr [8 x i8], ptr [[TMP24]], i64 [[TMP16]]
; CHECK-NEXT: store <vscale x 2 x i64> [[TMP21]], ptr [[TMP24]], align 8
; CHECK-NEXT: store <vscale x 2 x i64> [[TMP22]], ptr [[TMP26]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP2]]
; CHECK-NEXT: [[TMP27:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP27]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
;
entry:
br label %loop
loop:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
%gep.src.1 = getelementptr i64, ptr %src.1, i64 %iv
%gep.src.2 = getelementptr i64, ptr %src.2, i64 %iv
%l.1 = load i64, ptr %gep.src.1
%l.2 = load i64, ptr %gep.src.2
%add = add i64 %l.1, %l.2
%gep.dst.1 = getelementptr i64, ptr %dst.1, i64 %iv
%gep.dst.2 = getelementptr i64, ptr %dst.2, i64 %iv
store i64 %add, ptr %gep.dst.1
store i64 %add, ptr %gep.dst.2
%iv.next = add nsw i64 %iv, 1
%cmp10 = icmp ult i64 %iv.next, %n
br i1 %cmp10, label %loop, label %exit
exit:
ret void
}
;.
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
;.