llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^scalar.ph:" --version 5
; RUN: opt -p loop-vectorize -force-vector-width=2 -force-vector-interleave=1 -S %s | FileCheck --check-prefixes=VF2 %s
; RUN: opt -p loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S %s | FileCheck --check-prefixes=VF4 %s

target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-n32:64-S128-Fn32"
target triple = "arm64-apple-macosx"

define void @load_store_interleave_group(ptr noalias %data) {
; VF2-LABEL: define void @load_store_interleave_group(
; VF2-SAME: ptr noalias [[DATA:%.*]]) {
; VF2-NEXT:  [[ENTRY:.*:]]
; VF2-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VF2:       [[VECTOR_PH]]:
; VF2-NEXT:    br label %[[VECTOR_BODY:.*]]
; VF2:       [[VECTOR_BODY]]:
; VF2-NEXT:    [[TMP6:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; VF2-NEXT:    [[TMP8:%.*]] = shl nsw i64 [[TMP6]], 1
; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP8]]
; VF2-NEXT:    [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; VF2-NEXT:    store <2 x i64> [[WIDE_LOAD1]], ptr [[TMP5]], align 8
; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[TMP6]], 1
; VF2-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; VF2-NEXT:    br i1 [[TMP4]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; VF2:       [[MIDDLE_BLOCK]]:
; VF2-NEXT:    br i1 true, [[EXIT:label %.*]], label %[[SCALAR_PH]]
; VF2:       [[SCALAR_PH]]:
;
; VF4-LABEL: define void @load_store_interleave_group(
; VF4-SAME: ptr noalias [[DATA:%.*]]) {
; VF4-NEXT:  [[ENTRY:.*:]]
; VF4-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VF4:       [[VECTOR_PH]]:
; VF4-NEXT:    br label %[[VECTOR_BODY:.*]]
; VF4:       [[VECTOR_BODY]]:
; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; VF4-NEXT:    [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 1
; VF4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP1]]
; VF4-NEXT:    [[WIDE_VEC:%.*]] = load <8 x i64>, ptr [[TMP2]], align 8
; VF4-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <8 x i64> [[WIDE_VEC]], <8 x i64> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; VF4-NEXT:    [[STRIDED_VEC1:%.*]] = shufflevector <8 x i64> [[WIDE_VEC]], <8 x i64> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; VF4-NEXT:    [[TMP3:%.*]] = shufflevector <4 x i64> [[STRIDED_VEC]], <4 x i64> [[STRIDED_VEC1]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
; VF4-NEXT:    [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x i64> [[TMP3]], <8 x i64> poison, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
; VF4-NEXT:    store <8 x i64> [[INTERLEAVED_VEC]], ptr [[TMP2]], align 8
; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VF4-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; VF4-NEXT:    br i1 [[TMP4]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; VF4:       [[MIDDLE_BLOCK]]:
; VF4-NEXT:    br i1 true, [[EXIT:label %.*]], label %[[SCALAR_PH]]
; VF4:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %mul.2 = shl nsw i64 %iv, 1
  %data.0 = getelementptr inbounds i64, ptr %data, i64 %mul.2
  %l.0 = load i64, ptr %data.0, align 8
  store i64 %l.0, ptr %data.0, align 8
  %add.1 = or disjoint i64 %mul.2, 1
  %data.1 = getelementptr inbounds i64, ptr %data, i64 %add.1
  %l.1 = load i64, ptr %data.1, align 8
  store i64 %l.1, ptr %data.1, align 8
  %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
}

define void @load_store_interleave_group_different_objecs(ptr noalias %src, ptr noalias %dst) {
; VF2-LABEL: define void @load_store_interleave_group_different_objecs(
; VF2-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
; VF2-NEXT:  [[ENTRY:.*:]]
; VF2-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VF2:       [[VECTOR_PH]]:
; VF2-NEXT:    br label %[[VECTOR_BODY:.*]]
; VF2:       [[VECTOR_BODY]]:
; VF2-NEXT:    [[TMP6:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; VF2-NEXT:    [[TMP8:%.*]] = shl nsw i64 [[TMP6]], 1
; VF2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[SRC]], i64 [[TMP8]]
; VF2-NEXT:    [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP10]], align 8
; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[TMP8]]
; VF2-NEXT:    store <2 x i64> [[WIDE_LOAD1]], ptr [[TMP7]], align 8
; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[TMP6]], 1
; VF2-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; VF2-NEXT:    br i1 [[TMP5]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; VF2:       [[MIDDLE_BLOCK]]:
; VF2-NEXT:    br i1 true, [[EXIT:label %.*]], label %[[SCALAR_PH]]
; VF2:       [[SCALAR_PH]]:
;
; VF4-LABEL: define void @load_store_interleave_group_different_objecs(
; VF4-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
; VF4-NEXT:  [[ENTRY:.*:]]
; VF4-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VF4:       [[VECTOR_PH]]:
; VF4-NEXT:    br label %[[VECTOR_BODY:.*]]
; VF4:       [[VECTOR_BODY]]:
; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; VF4-NEXT:    [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 1
; VF4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[SRC]], i64 [[TMP1]]
; VF4-NEXT:    [[WIDE_VEC:%.*]] = load <8 x i64>, ptr [[TMP2]], align 8
; VF4-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <8 x i64> [[WIDE_VEC]], <8 x i64> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; VF4-NEXT:    [[STRIDED_VEC1:%.*]] = shufflevector <8 x i64> [[WIDE_VEC]], <8 x i64> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; VF4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[TMP1]]
; VF4-NEXT:    [[TMP4:%.*]] = shufflevector <4 x i64> [[STRIDED_VEC]], <4 x i64> [[STRIDED_VEC1]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
; VF4-NEXT:    [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x i64> [[TMP4]], <8 x i64> poison, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
; VF4-NEXT:    store <8 x i64> [[INTERLEAVED_VEC]], ptr [[TMP3]], align 8
; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VF4-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; VF4-NEXT:    br i1 [[TMP5]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; VF4:       [[MIDDLE_BLOCK]]:
; VF4-NEXT:    br i1 true, [[EXIT:label %.*]], label %[[SCALAR_PH]]
; VF4:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %mul.2 = shl nsw i64 %iv, 1
  %src.0 = getelementptr inbounds i64, ptr %src, i64 %mul.2
  %l.0 = load i64, ptr %src.0, align 8
  %dst.0 = getelementptr inbounds i64, ptr %dst, i64 %mul.2
  store i64 %l.0, ptr %dst.0, align 8
  %add.1 = or disjoint i64 %mul.2, 1
  %src.1 = getelementptr inbounds i64, ptr %src, i64 %add.1
  %l.1 = load i64, ptr %src.1, align 8
  %dst.1 = getelementptr inbounds i64, ptr %dst, i64 %add.1
  store i64 %l.1, ptr %dst.1, align 8
  %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
}

define void @single_wide_load_store_interleave_group(ptr noalias %src, ptr noalias %dst) {
; VF2-LABEL: define void @single_wide_load_store_interleave_group(
; VF2-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
; VF2-NEXT:  [[ENTRY:.*:]]
; VF2-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VF2:       [[VECTOR_PH]]:
; VF2-NEXT:    br label %[[VECTOR_BODY:.*]]
; VF2:       [[VECTOR_BODY]]:
; VF2-NEXT:    [[TMP7:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; VF2-NEXT:    [[TMP9:%.*]] = shl nsw i64 [[TMP7]], 1
; VF2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[SRC]], i64 [[TMP7]]
; VF2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; VF2-NEXT:    [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP3]], align 8
; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[TMP9]]
; VF2-NEXT:    [[TMP10:%.*]] = shufflevector <2 x i64> [[WIDE_LOAD1]], <2 x i64> [[WIDE_LOAD1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
; VF2-NEXT:    [[INTERLEAVED_VEC2:%.*]] = shufflevector <4 x i64> [[TMP10]], <4 x i64> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
; VF2-NEXT:    store <4 x i64> [[INTERLEAVED_VEC2]], ptr [[TMP8]], align 8
; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[TMP7]], 2
; VF2-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; VF2-NEXT:    br i1 [[TMP6]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; VF2:       [[MIDDLE_BLOCK]]:
; VF2-NEXT:    br i1 true, [[EXIT:label %.*]], label %[[SCALAR_PH]]
; VF2:       [[SCALAR_PH]]:
;
; VF4-LABEL: define void @single_wide_load_store_interleave_group(
; VF4-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
; VF4-NEXT:  [[ENTRY:.*:]]
; VF4-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VF4:       [[VECTOR_PH]]:
; VF4-NEXT:    br label %[[VECTOR_BODY:.*]]
; VF4:       [[VECTOR_BODY]]:
; VF4-NEXT:    [[TMP7:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; VF4-NEXT:    [[TMP9:%.*]] = shl nsw i64 [[TMP7]], 1
; VF4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[SRC]], i64 [[TMP7]]
; VF4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; VF4-NEXT:    [[WIDE_LOAD1:%.*]] = load <4 x i64>, ptr [[TMP3]], align 8
; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[TMP9]]
; VF4-NEXT:    [[TMP10:%.*]] = shufflevector <4 x i64> [[WIDE_LOAD1]], <4 x i64> [[WIDE_LOAD1]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
; VF4-NEXT:    [[INTERLEAVED_VEC2:%.*]] = shufflevector <8 x i64> [[TMP10]], <8 x i64> poison, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
; VF4-NEXT:    store <8 x i64> [[INTERLEAVED_VEC2]], ptr [[TMP8]], align 8
; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[TMP7]], 4
; VF4-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; VF4-NEXT:    br i1 [[TMP5]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; VF4:       [[MIDDLE_BLOCK]]:
; VF4-NEXT:    br i1 true, [[EXIT:label %.*]], label %[[SCALAR_PH]]
; VF4:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %mul.2 = shl nsw i64 %iv, 1
  %src.0 = getelementptr inbounds i64, ptr %src, i64 %iv
  %l.0 = load i64, ptr %src.0, align 8
  %dst.0 = getelementptr inbounds i64, ptr %dst, i64 %mul.2
  store i64 %l.0, ptr %dst.0, align 8
  %add.1 = or disjoint i64 %mul.2, 1
  %dst.1 = getelementptr inbounds i64, ptr %dst, i64 %add.1
  store i64 %l.0, ptr %dst.1, align 8
  %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
}