[AArch64 NEON] Try to generate CONCAT_VECTOR when lowering BUILD_VECTOR or SHUFFLE_VECTOR.

Replace r199791.

llvm-svn: 200180

lib/Target/AArch64/AArch64ISelLowering.cpp

@@ -4153,22 +4153,76 @@ AArch64TargetLowering::isFMAFasterThanFMulAndFAdd(EVT VT) const {
 
   return false;
 }
+// Check whether a shuffle_vector could be presented as concat_vector.
+bool AArch64TargetLowering::isConcatVector(SDValue Op, SelectionDAG &DAG,
+                                           SDValue V0, SDValue V1,
+                                           const int *Mask,
+                                           SDValue &Res) const {
+  SDLoc DL(Op);
+  EVT VT = Op.getValueType();
+  if (VT.getSizeInBits() != 128)
+    return false;
+  if (VT.getVectorElementType() != V0.getValueType().getVectorElementType() ||
+      VT.getVectorElementType() != V1.getValueType().getVectorElementType())
+    return false;
 
-// Check whether a Build Vector could be presented as Shuffle Vector. If yes,
+  unsigned NumElts = VT.getVectorNumElements();
-// try to call LowerVECTOR_SHUFFLE to lower it.
+  bool isContactVector = true;
+  bool splitV0 = false;
+  if (V0.getValueType().getSizeInBits() == 128)
+    splitV0 = true;
+
+  for (int I = 0, E = NumElts / 2; I != E; I++) {
+    if (Mask[I] != I) {
+      isContactVector = false;
+      break;
+    }
+  }
+
+  if (isContactVector) {
+    int offset = NumElts / 2;
+    for (int I = NumElts / 2, E = NumElts; I != E; I++) {
+      if (Mask[I] != I + splitV0 * offset) {
+        isContactVector = false;
+        break;
+      }
+    }
+  }
+
+  if (isContactVector) {
+    EVT CastVT = EVT::getVectorVT(*DAG.getContext(), VT.getVectorElementType(),
+                                  NumElts / 2);
+    if (splitV0) {
+      V0 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, CastVT, V0,
+                       DAG.getConstant(0, MVT::i64));
+    }
+    if (V1.getValueType().getSizeInBits() == 128) {
+      V1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, CastVT, V1,
+                       DAG.getConstant(0, MVT::i64));
+    }
+    Res = DAG.getNode(ISD::CONCAT_VECTORS, DL, VT, V0, V1);
+    return true;
+  }
+  return false;
+}
+
+// Check whether a Build Vector could be presented as Shuffle Vector.
+// This Shuffle Vector maybe not legalized, so the length of its operand and
+// the length of result may not equal.
 bool AArch64TargetLowering::isKnownShuffleVector(SDValue Op, SelectionDAG &DAG,
-                                                 SDValue &Res) const {
+                                                 SDValue &V0, SDValue &V1,
+                                                 int *Mask) const {
   SDLoc DL(Op);
   EVT VT = Op.getValueType();
   unsigned NumElts = VT.getVectorNumElements();
   unsigned V0NumElts = 0;
-  int Mask[16];
-  SDValue V0, V1;
 
   // Check if all elements are extracted from less than 3 vectors.
   for (unsigned i = 0; i < NumElts; ++i) {
     SDValue Elt = Op.getOperand(i);
-    if (Elt.getOpcode() != ISD::EXTRACT_VECTOR_ELT)
+    if (Elt.getOpcode() != ISD::EXTRACT_VECTOR_ELT ||
+        Elt.getOperand(0).getValueType().getVectorElementType() !=
+            VT.getVectorElementType())
       return false;
 
     if (V0.getNode() == 0) {
@@ -4189,25 +4243,7 @@ bool AArch64TargetLowering::isKnownShuffleVector(SDValue Op, SelectionDAG &DAG,
       return false;
     }
   }
-
+  return true;
-  if (!V1.getNode() && V0NumElts == NumElts * 2) {
-    V1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, V0,
-                     DAG.getConstant(NumElts, MVT::i64));
-    V0 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, V0,
-                     DAG.getConstant(0, MVT::i64));
-    V0NumElts = V0.getValueType().getVectorNumElements();
-  }
-
-  if (V1.getNode() && NumElts == V0NumElts &&
-      V0NumElts == V1.getValueType().getVectorNumElements()) {
-    SDValue Shuffle = DAG.getVectorShuffle(VT, DL, V0, V1, Mask);
-    if(Shuffle.getOpcode() != ISD::VECTOR_SHUFFLE)
-      Res = Shuffle;
-    else
-      Res = LowerVECTOR_SHUFFLE(Shuffle, DAG);
-    return true;
-  } else
-    return false;
 }
 
 // If this is a case we can't handle, return null and let the default
@@ -4413,9 +4449,31 @@ AArch64TargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
     return SDValue();
 
   // Try to lower this in lowering ShuffleVector way.
-  SDValue Shuf;
+  SDValue V0, V1;
-  if (isKnownShuffleVector(Op, DAG, Shuf))
+  int Mask[16];
-    return Shuf;
+  if (isKnownShuffleVector(Op, DAG, V0, V1, Mask)) {
+    unsigned V0NumElts = V0.getValueType().getVectorNumElements();
+    if (!V1.getNode() && V0NumElts == NumElts * 2) {
+      V1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, V0,
+                       DAG.getConstant(NumElts, MVT::i64));
+      V0 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, V0,
+                       DAG.getConstant(0, MVT::i64));
+      V0NumElts = V0.getValueType().getVectorNumElements();
+    }
+
+    if (V1.getNode() && NumElts == V0NumElts &&
+        V0NumElts == V1.getValueType().getVectorNumElements()) {
+      SDValue Shuffle = DAG.getVectorShuffle(VT, DL, V0, V1, Mask);
+      if (Shuffle.getOpcode() != ISD::VECTOR_SHUFFLE)
+        return Shuffle;
+      else
+        return LowerVECTOR_SHUFFLE(Shuffle, DAG);
+    } else {
+      SDValue Res;
+      if (isConcatVector(Op, DAG, V0, V1, Mask, Res))
+        return Res;
+    }
+  }
 
   // If all else fails, just use a sequence of INSERT_VECTOR_ELT when we
   // know the default expansion would otherwise fall back on something even
@@ -4601,6 +4659,10 @@ AArch64TargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
       return DAG.getNode(ISDNo, dl, VT, V1, V2);
   }
 
+  SDValue Res;
+  if (isConcatVector(Op, DAG, V1, V2, &ShuffleMask[0], Res))
+    return Res;
+
   // If the element of shuffle mask are all the same constant, we can
   // transform it into either NEON_VDUP or NEON_VDUPLANE
   if (ShuffleVectorSDNode::isSplatMask(&ShuffleMask[0], VT)) {

lib/Target/AArch64/AArch64ISelLowering.h

@@ -232,7 +232,11 @@ public:
                           SDLoc dl, SelectionDAG &DAG,
                           SmallVectorImpl<SDValue> &InVals) const;
 
-  bool isKnownShuffleVector(SDValue Op, SelectionDAG &DAG, SDValue &Res) const;
+  bool isConcatVector(SDValue Op, SelectionDAG &DAG, SDValue V0, SDValue V1,
+                      const int *Mask, SDValue &Res) const;
+
+  bool isKnownShuffleVector(SDValue Op, SelectionDAG &DAG, SDValue &V0,
+                            SDValue &V1, int *Mask) const;
 
   SDValue LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
                             const AArch64Subtarget *ST) const;

test/CodeGen/AArch64/neon-copy.ll

@@ -1021,6 +1021,276 @@ entry:
   ret <2 x i32> %h
 }
 
+define <16 x i8> @test_concat_v16i8_v16i8_v16i8(<16 x i8> %x, <16 x i8> %y) #0 {
+; CHECK-LABEL: test_concat_v16i8_v16i8_v16i8:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecinit30 = shufflevector <16 x i8> %x, <16 x i8> %y, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
+  ret <16 x i8> %vecinit30
+}
+
+define <16 x i8> @test_concat_v16i8_v8i8_v16i8(<8 x i8> %x, <16 x i8> %y) #0 {
+; CHECK-LABEL: test_concat_v16i8_v8i8_v16i8:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <8 x i8> %x, i32 0
+  %vecinit = insertelement <16 x i8> undef, i8 %vecext, i32 0
+  %vecext1 = extractelement <8 x i8> %x, i32 1
+  %vecinit2 = insertelement <16 x i8> %vecinit, i8 %vecext1, i32 1
+  %vecext3 = extractelement <8 x i8> %x, i32 2
+  %vecinit4 = insertelement <16 x i8> %vecinit2, i8 %vecext3, i32 2
+  %vecext5 = extractelement <8 x i8> %x, i32 3
+  %vecinit6 = insertelement <16 x i8> %vecinit4, i8 %vecext5, i32 3
+  %vecext7 = extractelement <8 x i8> %x, i32 4
+  %vecinit8 = insertelement <16 x i8> %vecinit6, i8 %vecext7, i32 4
+  %vecext9 = extractelement <8 x i8> %x, i32 5
+  %vecinit10 = insertelement <16 x i8> %vecinit8, i8 %vecext9, i32 5
+  %vecext11 = extractelement <8 x i8> %x, i32 6
+  %vecinit12 = insertelement <16 x i8> %vecinit10, i8 %vecext11, i32 6
+  %vecext13 = extractelement <8 x i8> %x, i32 7
+  %vecinit14 = insertelement <16 x i8> %vecinit12, i8 %vecext13, i32 7
+  %vecinit30 = shufflevector <16 x i8> %vecinit14, <16 x i8> %y, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
+  ret <16 x i8> %vecinit30
+}
+
+define <16 x i8> @test_concat_v16i8_v16i8_v8i8(<16 x i8> %x, <8 x i8> %y) #0 {
+; CHECK-LABEL: test_concat_v16i8_v16i8_v8i8:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <16 x i8> %x, i32 0
+  %vecinit = insertelement <16 x i8> undef, i8 %vecext, i32 0
+  %vecext1 = extractelement <16 x i8> %x, i32 1
+  %vecinit2 = insertelement <16 x i8> %vecinit, i8 %vecext1, i32 1
+  %vecext3 = extractelement <16 x i8> %x, i32 2
+  %vecinit4 = insertelement <16 x i8> %vecinit2, i8 %vecext3, i32 2
+  %vecext5 = extractelement <16 x i8> %x, i32 3
+  %vecinit6 = insertelement <16 x i8> %vecinit4, i8 %vecext5, i32 3
+  %vecext7 = extractelement <16 x i8> %x, i32 4
+  %vecinit8 = insertelement <16 x i8> %vecinit6, i8 %vecext7, i32 4
+  %vecext9 = extractelement <16 x i8> %x, i32 5
+  %vecinit10 = insertelement <16 x i8> %vecinit8, i8 %vecext9, i32 5
+  %vecext11 = extractelement <16 x i8> %x, i32 6
+  %vecinit12 = insertelement <16 x i8> %vecinit10, i8 %vecext11, i32 6
+  %vecext13 = extractelement <16 x i8> %x, i32 7
+  %vecinit14 = insertelement <16 x i8> %vecinit12, i8 %vecext13, i32 7
+  %vecext15 = extractelement <8 x i8> %y, i32 0
+  %vecinit16 = insertelement <16 x i8> %vecinit14, i8 %vecext15, i32 8
+  %vecext17 = extractelement <8 x i8> %y, i32 1
+  %vecinit18 = insertelement <16 x i8> %vecinit16, i8 %vecext17, i32 9
+  %vecext19 = extractelement <8 x i8> %y, i32 2
+  %vecinit20 = insertelement <16 x i8> %vecinit18, i8 %vecext19, i32 10
+  %vecext21 = extractelement <8 x i8> %y, i32 3
+  %vecinit22 = insertelement <16 x i8> %vecinit20, i8 %vecext21, i32 11
+  %vecext23 = extractelement <8 x i8> %y, i32 4
+  %vecinit24 = insertelement <16 x i8> %vecinit22, i8 %vecext23, i32 12
+  %vecext25 = extractelement <8 x i8> %y, i32 5
+  %vecinit26 = insertelement <16 x i8> %vecinit24, i8 %vecext25, i32 13
+  %vecext27 = extractelement <8 x i8> %y, i32 6
+  %vecinit28 = insertelement <16 x i8> %vecinit26, i8 %vecext27, i32 14
+  %vecext29 = extractelement <8 x i8> %y, i32 7
+  %vecinit30 = insertelement <16 x i8> %vecinit28, i8 %vecext29, i32 15
+  ret <16 x i8> %vecinit30
+}
+
+define <16 x i8> @test_concat_v16i8_v8i8_v8i8(<8 x i8> %x, <8 x i8> %y) #0 {
+; CHECK-LABEL: test_concat_v16i8_v8i8_v8i8:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <8 x i8> %x, i32 0
+  %vecinit = insertelement <16 x i8> undef, i8 %vecext, i32 0
+  %vecext1 = extractelement <8 x i8> %x, i32 1
+  %vecinit2 = insertelement <16 x i8> %vecinit, i8 %vecext1, i32 1
+  %vecext3 = extractelement <8 x i8> %x, i32 2
+  %vecinit4 = insertelement <16 x i8> %vecinit2, i8 %vecext3, i32 2
+  %vecext5 = extractelement <8 x i8> %x, i32 3
+  %vecinit6 = insertelement <16 x i8> %vecinit4, i8 %vecext5, i32 3
+  %vecext7 = extractelement <8 x i8> %x, i32 4
+  %vecinit8 = insertelement <16 x i8> %vecinit6, i8 %vecext7, i32 4
+  %vecext9 = extractelement <8 x i8> %x, i32 5
+  %vecinit10 = insertelement <16 x i8> %vecinit8, i8 %vecext9, i32 5
+  %vecext11 = extractelement <8 x i8> %x, i32 6
+  %vecinit12 = insertelement <16 x i8> %vecinit10, i8 %vecext11, i32 6
+  %vecext13 = extractelement <8 x i8> %x, i32 7
+  %vecinit14 = insertelement <16 x i8> %vecinit12, i8 %vecext13, i32 7
+  %vecext15 = extractelement <8 x i8> %y, i32 0
+  %vecinit16 = insertelement <16 x i8> %vecinit14, i8 %vecext15, i32 8
+  %vecext17 = extractelement <8 x i8> %y, i32 1
+  %vecinit18 = insertelement <16 x i8> %vecinit16, i8 %vecext17, i32 9
+  %vecext19 = extractelement <8 x i8> %y, i32 2
+  %vecinit20 = insertelement <16 x i8> %vecinit18, i8 %vecext19, i32 10
+  %vecext21 = extractelement <8 x i8> %y, i32 3
+  %vecinit22 = insertelement <16 x i8> %vecinit20, i8 %vecext21, i32 11
+  %vecext23 = extractelement <8 x i8> %y, i32 4
+  %vecinit24 = insertelement <16 x i8> %vecinit22, i8 %vecext23, i32 12
+  %vecext25 = extractelement <8 x i8> %y, i32 5
+  %vecinit26 = insertelement <16 x i8> %vecinit24, i8 %vecext25, i32 13
+  %vecext27 = extractelement <8 x i8> %y, i32 6
+  %vecinit28 = insertelement <16 x i8> %vecinit26, i8 %vecext27, i32 14
+  %vecext29 = extractelement <8 x i8> %y, i32 7
+  %vecinit30 = insertelement <16 x i8> %vecinit28, i8 %vecext29, i32 15
+  ret <16 x i8> %vecinit30
+}
+
+define <8 x i16> @test_concat_v8i16_v8i16_v8i16(<8 x i16> %x, <8 x i16> %y) #0 {
+; CHECK-LABEL: test_concat_v8i16_v8i16_v8i16:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecinit14 = shufflevector <8 x i16> %x, <8 x i16> %y, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
+  ret <8 x i16> %vecinit14
+}
+
+define <8 x i16> @test_concat_v8i16_v4i16_v8i16(<4 x i16> %x, <8 x i16> %y) #0 {
+; CHECK-LABEL: test_concat_v8i16_v4i16_v8i16:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <4 x i16> %x, i32 0
+  %vecinit = insertelement <8 x i16> undef, i16 %vecext, i32 0
+  %vecext1 = extractelement <4 x i16> %x, i32 1
+  %vecinit2 = insertelement <8 x i16> %vecinit, i16 %vecext1, i32 1
+  %vecext3 = extractelement <4 x i16> %x, i32 2
+  %vecinit4 = insertelement <8 x i16> %vecinit2, i16 %vecext3, i32 2
+  %vecext5 = extractelement <4 x i16> %x, i32 3
+  %vecinit6 = insertelement <8 x i16> %vecinit4, i16 %vecext5, i32 3
+  %vecinit14 = shufflevector <8 x i16> %vecinit6, <8 x i16> %y, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
+  ret <8 x i16> %vecinit14
+}
+
+define <8 x i16> @test_concat_v8i16_v8i16_v4i16(<8 x i16> %x, <4 x i16> %y) #0 {
+; CHECK-LABEL: test_concat_v8i16_v8i16_v4i16:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <8 x i16> %x, i32 0
+  %vecinit = insertelement <8 x i16> undef, i16 %vecext, i32 0
+  %vecext1 = extractelement <8 x i16> %x, i32 1
+  %vecinit2 = insertelement <8 x i16> %vecinit, i16 %vecext1, i32 1
+  %vecext3 = extractelement <8 x i16> %x, i32 2
+  %vecinit4 = insertelement <8 x i16> %vecinit2, i16 %vecext3, i32 2
+  %vecext5 = extractelement <8 x i16> %x, i32 3
+  %vecinit6 = insertelement <8 x i16> %vecinit4, i16 %vecext5, i32 3
+  %vecext7 = extractelement <4 x i16> %y, i32 0
+  %vecinit8 = insertelement <8 x i16> %vecinit6, i16 %vecext7, i32 4
+  %vecext9 = extractelement <4 x i16> %y, i32 1
+  %vecinit10 = insertelement <8 x i16> %vecinit8, i16 %vecext9, i32 5
+  %vecext11 = extractelement <4 x i16> %y, i32 2
+  %vecinit12 = insertelement <8 x i16> %vecinit10, i16 %vecext11, i32 6
+  %vecext13 = extractelement <4 x i16> %y, i32 3
+  %vecinit14 = insertelement <8 x i16> %vecinit12, i16 %vecext13, i32 7
+  ret <8 x i16> %vecinit14
+}
+
+define <8 x i16> @test_concat_v8i16_v4i16_v4i16(<4 x i16> %x, <4 x i16> %y) #0 {
+; CHECK-LABEL: test_concat_v8i16_v4i16_v4i16:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <4 x i16> %x, i32 0
+  %vecinit = insertelement <8 x i16> undef, i16 %vecext, i32 0
+  %vecext1 = extractelement <4 x i16> %x, i32 1
+  %vecinit2 = insertelement <8 x i16> %vecinit, i16 %vecext1, i32 1
+  %vecext3 = extractelement <4 x i16> %x, i32 2
+  %vecinit4 = insertelement <8 x i16> %vecinit2, i16 %vecext3, i32 2
+  %vecext5 = extractelement <4 x i16> %x, i32 3
+  %vecinit6 = insertelement <8 x i16> %vecinit4, i16 %vecext5, i32 3
+  %vecext7 = extractelement <4 x i16> %y, i32 0
+  %vecinit8 = insertelement <8 x i16> %vecinit6, i16 %vecext7, i32 4
+  %vecext9 = extractelement <4 x i16> %y, i32 1
+  %vecinit10 = insertelement <8 x i16> %vecinit8, i16 %vecext9, i32 5
+  %vecext11 = extractelement <4 x i16> %y, i32 2
+  %vecinit12 = insertelement <8 x i16> %vecinit10, i16 %vecext11, i32 6
+  %vecext13 = extractelement <4 x i16> %y, i32 3
+  %vecinit14 = insertelement <8 x i16> %vecinit12, i16 %vecext13, i32 7
+  ret <8 x i16> %vecinit14
+}
+
+define <4 x i32> @test_concat_v4i32_v4i32_v4i32(<4 x i32> %x, <4 x i32> %y) #0 {
+; CHECK-LABEL: test_concat_v4i32_v4i32_v4i32:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecinit6 = shufflevector <4 x i32> %x, <4 x i32> %y, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+  ret <4 x i32> %vecinit6
+}
+
+define <4 x i32> @test_concat_v4i32_v2i32_v4i32(<2 x i32> %x, <4 x i32> %y) #0 {
+; CHECK-LABEL: test_concat_v4i32_v2i32_v4i32:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <2 x i32> %x, i32 0
+  %vecinit = insertelement <4 x i32> undef, i32 %vecext, i32 0
+  %vecext1 = extractelement <2 x i32> %x, i32 1
+  %vecinit2 = insertelement <4 x i32> %vecinit, i32 %vecext1, i32 1
+  %vecinit6 = shufflevector <4 x i32> %vecinit2, <4 x i32> %y, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+  ret <4 x i32> %vecinit6
+}
+
+define <4 x i32> @test_concat_v4i32_v4i32_v2i32(<4 x i32> %x, <2 x i32> %y) #0 {
+; CHECK-LABEL: test_concat_v4i32_v4i32_v2i32:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <4 x i32> %x, i32 0
+  %vecinit = insertelement <4 x i32> undef, i32 %vecext, i32 0
+  %vecext1 = extractelement <4 x i32> %x, i32 1
+  %vecinit2 = insertelement <4 x i32> %vecinit, i32 %vecext1, i32 1
+  %vecext3 = extractelement <2 x i32> %y, i32 0
+  %vecinit4 = insertelement <4 x i32> %vecinit2, i32 %vecext3, i32 2
+  %vecext5 = extractelement <2 x i32> %y, i32 1
+  %vecinit6 = insertelement <4 x i32> %vecinit4, i32 %vecext5, i32 3
+  ret <4 x i32> %vecinit6
+}
+
+define <4 x i32> @test_concat_v4i32_v2i32_v2i32(<2 x i32> %x, <2 x i32> %y) #0 {
+; CHECK-LABEL: test_concat_v4i32_v2i32_v2i32:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <2 x i32> %x, i32 0
+  %vecinit = insertelement <4 x i32> undef, i32 %vecext, i32 0
+  %vecext1 = extractelement <2 x i32> %x, i32 1
+  %vecinit2 = insertelement <4 x i32> %vecinit, i32 %vecext1, i32 1
+  %vecext3 = extractelement <2 x i32> %y, i32 0
+  %vecinit4 = insertelement <4 x i32> %vecinit2, i32 %vecext3, i32 2
+  %vecext5 = extractelement <2 x i32> %y, i32 1
+  %vecinit6 = insertelement <4 x i32> %vecinit4, i32 %vecext5, i32 3
+  ret <4 x i32> %vecinit6
+}
+
+define <2 x i64> @test_concat_v2i64_v2i64_v2i64(<2 x i64> %x, <2 x i64> %y) #0 {
+; CHECK-LABEL: test_concat_v2i64_v2i64_v2i64:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecinit2 = shufflevector <2 x i64> %x, <2 x i64> %y, <2 x i32> <i32 0, i32 2>
+  ret <2 x i64> %vecinit2
+}
+
+define <2 x i64> @test_concat_v2i64_v1i64_v2i64(<1 x i64> %x, <2 x i64> %y) #0 {
+; CHECK-LABEL: test_concat_v2i64_v1i64_v2i64:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <1 x i64> %x, i32 0
+  %vecinit = insertelement <2 x i64> undef, i64 %vecext, i32 0
+  %vecinit2 = shufflevector <2 x i64> %vecinit, <2 x i64> %y, <2 x i32> <i32 0, i32 2>
+  ret <2 x i64> %vecinit2
+}
+
+define <2 x i64> @test_concat_v2i64_v2i64_v1i64(<2 x i64> %x, <1 x i64> %y) #0 {
+; CHECK-LABEL: test_concat_v2i64_v2i64_v1i64:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <2 x i64> %x, i32 0
+  %vecinit = insertelement <2 x i64> undef, i64 %vecext, i32 0
+  %vecext1 = extractelement <1 x i64> %y, i32 0
+  %vecinit2 = insertelement <2 x i64> %vecinit, i64 %vecext1, i32 1
+  ret <2 x i64> %vecinit2
+}
+
+define <2 x i64> @test_concat_v2i64_v1i64_v1i64(<1 x i64> %x, <1 x i64> %y) #0 {
+; CHECK-LABEL: test_concat_v2i64_v1i64_v1i64:
+; CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
+entry:
+  %vecext = extractelement <1 x i64> %x, i32 0
+  %vecinit = insertelement <2 x i64> undef, i64 %vecext, i32 0
+  %vecext1 = extractelement <1 x i64> %y, i32 0
+  %vecinit2 = insertelement <2 x i64> %vecinit, i64 %vecext1, i32 1
+  ret <2 x i64> %vecinit2
+}
+
 declare <1 x i16> @llvm.aarch64.neon.vsqadd.v1i16(<1 x i16>, <1 x i16>)
 
 define <1 x i16> @test_copy_FPR16_FPR16(<1 x i16> %a, <1 x i16> %b) {