blob: 0ccc743ecfb3cf3241a2494ffbea21362740e5b8 [file] [edit]
; 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 -p loop-vectorize -enable-interleaved-mem-accesses -force-vector-width=4 -S %s | FileCheck %s
define void @dep_between_interleaved_store_and_load_with_runtime_checks(ptr %a, ptr %b, i64 %off) {
; CHECK-LABEL: define void @dep_between_interleaved_store_and_load_with_runtime_checks(
; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[OFF:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[B_OFF:%.*]] = getelementptr i8, ptr [[B]], i64 [[OFF]]
; CHECK-NEXT: br label %[[VECTOR_MEMCHECK:.*]]
; CHECK: [[VECTOR_MEMCHECK]]:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 800
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[B]], i64 400
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[OFF]], 400
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP2]]
; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[A]], [[TMP1]]
; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[B]], [[TMP0]]
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: [[BOUND010:%.*]] = icmp ult ptr [[A]], [[TMP3]]
; CHECK-NEXT: [[BOUND111:%.*]] = icmp ult ptr [[B_OFF]], [[TMP0]]
; CHECK-NEXT: [[FOUND_CONFLICT12:%.*]] = and i1 [[BOUND010]], [[BOUND111]]
; CHECK-NEXT: [[CONFLICT_RDX:%.*]] = or i1 [[FOUND_CONFLICT]], [[FOUND_CONFLICT12]]
; CHECK-NEXT: [[BOUND013:%.*]] = icmp ult ptr [[B]], [[TMP3]]
; CHECK-NEXT: [[BOUND114:%.*]] = icmp ult ptr [[B_OFF]], [[TMP1]]
; CHECK-NEXT: [[FOUND_CONFLICT15:%.*]] = and i1 [[BOUND013]], [[BOUND114]]
; CHECK-NEXT: [[CONFLICT_RDX16:%.*]] = or i1 [[CONFLICT_RDX]], [[FOUND_CONFLICT15]]
; CHECK-NEXT: br i1 [[CONFLICT_RDX16]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; 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 [2 x i32], ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP4]], align 4, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT: store <8 x i32> <i32 1, i32 3, i32 1, i32 3, i32 1, i32 3, i32 1, i32 3>, ptr [[TMP4]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
; CHECK-NEXT: store <4 x i32> [[STRIDED_VEC]], ptr [[TMP5]], align 4, !alias.scope [[META6:![0-9]+]], !noalias [[META7:![0-9]+]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[B_OFF]], i64 [[INDEX]]
; CHECK-NEXT: store <4 x i32> zeroinitializer, ptr [[TMP6]], align 4, !alias.scope [[META7]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], 96
; CHECK-NEXT: br i1 [[TMP7]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
;
entry:
%b.off = getelementptr i8, ptr %b, i64 %off
br label %loop
loop:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
%gep.a.0 = getelementptr inbounds [2 x i32], ptr %a, i64 %iv
store i32 1, ptr %gep.a.0, align 4
%l.a.0 = load i32, ptr %gep.a.0, align 4
%gep.a.1 = getelementptr inbounds [2 x i32], ptr %a, i64 %iv, i64 1
store i32 3, ptr %gep.a.1, align 4
%gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
store i32 %l.a.0, ptr %gep.b, align 4
%gep.b.off = getelementptr inbounds i32, ptr %b.off, i64 %iv
store i32 0, ptr %gep.b.off, align 4
%iv.next = add nuw nsw i64 %iv, 1
%ec = icmp eq i64 %iv.next, 100
br i1 %ec, label %exit, label %loop
exit:
ret void
}