Type legalizer for masked gather and scatter intrinsics.

Full type legalizer that works with all vectors length - from 2 to 16, (i32, i64, float, double).

This intrinsic, for example
void @llvm.masked.scatter.v2f32(<2 x float>%data , <2 x float*>%ptrs , i32 align , <2 x i1>%mask )
requires type widening for data and type promotion for mask.

Differential Revision: http://reviews.llvm.org/D13633

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@255629 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/llvm/CodeGen/SelectionDAGNodes.h b/include/llvm/CodeGen/SelectionDAGNodes.h
index 548549ab1353404921c0f18e261e41c7493009bb..23816bde07c0de9e12a3e45b2ff0acace8f79560 100644 (file)
--- a/include/llvm/CodeGen/SelectionDAGNodes.h
+++ b/include/llvm/CodeGen/SelectionDAGNodes.h
@@ -2122,12 +2122,13 @@ public:
     : MaskedGatherScatterSDNode(ISD::MGATHER, Order, dl, Operands, VTs, MemVT,
                                 MMO) {
     assert(getValue().getValueType() == getValueType(0) &&
-           "Incompatible type of the PathThru value in MaskedGatherSDNode");
+           "Incompatible type of the PassThru value in MaskedGatherSDNode");
     assert(getMask().getValueType().getVectorNumElements() ==
-               getValueType(0).getVectorNumElements() &&
-           "Vector width mismatch between mask and data");
-    assert(getMask().getValueType().getScalarType() == MVT::i1 &&
+           getValueType(0).getVectorNumElements() &&
            "Vector width mismatch between mask and data");
+    assert(getIndex().getValueType().getVectorNumElements() ==
+           getValueType(0).getVectorNumElements() &&
+           "Vector width mismatch between index and data");
   }
 
   static bool classof(const SDNode *N) {
@@ -2143,13 +2144,14 @@ public:
   friend class SelectionDAG;
   MaskedScatterSDNode(unsigned Order, DebugLoc dl,ArrayRef<SDValue> Operands,
                       SDVTList VTs, EVT MemVT, MachineMemOperand *MMO)
-      : MaskedGatherScatterSDNode(ISD::MSCATTER, Order, dl, Operands, VTs,
-                                  MemVT, MMO) {
+    : MaskedGatherScatterSDNode(ISD::MSCATTER, Order, dl, Operands, VTs, MemVT,
+                                MMO) {
     assert(getMask().getValueType().getVectorNumElements() ==
-               getValue().getValueType().getVectorNumElements() &&
-           "Vector width mismatch between mask and data");
-    assert(getMask().getValueType().getScalarType() == MVT::i1 &&
+           getValue().getValueType().getVectorNumElements() &&
            "Vector width mismatch between mask and data");
+    assert(getIndex().getValueType().getVectorNumElements() ==
+           getValue().getValueType().getVectorNumElements() &&
+           "Vector width mismatch between index and data");
   }
 
   static bool classof(const SDNode *N) {

diff --git a/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp b/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp
index 2cfcf77b17a58ac7434528630c1a493555e68359..63c9cc52871032b977cf05e3b64143a865d7eeae 100644 (file)
--- a/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp
+++ b/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp
@@ -66,8 +66,11 @@ void DAGTypeLegalizer::PromoteIntegerResult(SDNode *N, unsigned ResNo) {
   case ISD::CTTZ:        Res = PromoteIntRes_CTTZ(N); break;
   case ISD::EXTRACT_VECTOR_ELT:
                          Res = PromoteIntRes_EXTRACT_VECTOR_ELT(N); break;
-  case ISD::LOAD:        Res = PromoteIntRes_LOAD(cast<LoadSDNode>(N));break;
-  case ISD::MLOAD:       Res = PromoteIntRes_MLOAD(cast<MaskedLoadSDNode>(N));break;
+  case ISD::LOAD:        Res = PromoteIntRes_LOAD(cast<LoadSDNode>(N)); break;
+  case ISD::MLOAD:       Res = PromoteIntRes_MLOAD(cast<MaskedLoadSDNode>(N));
+    break;
+  case ISD::MGATHER:     Res = PromoteIntRes_MGATHER(cast<MaskedGatherSDNode>(N));
+    break;
   case ISD::SELECT:      Res = PromoteIntRes_SELECT(N); break;
   case ISD::VSELECT:     Res = PromoteIntRes_VSELECT(N); break;
   case ISD::SELECT_CC:   Res = PromoteIntRes_SELECT_CC(N); break;
@@ -181,7 +184,7 @@ SDValue DAGTypeLegalizer::PromoteIntRes_Atomic0(AtomicSDNode *N) {
                               N->getChain(), N->getBasePtr(),
                               N->getMemOperand(), N->getOrdering(),
                               N->getSynchScope());
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
   return Res;
@@ -194,7 +197,7 @@ SDValue DAGTypeLegalizer::PromoteIntRes_Atomic1(AtomicSDNode *N) {
                               N->getChain(), N->getBasePtr(),
                               Op2, N->getMemOperand(), N->getOrdering(),
                               N->getSynchScope());
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
   return Res;
@@ -479,7 +482,7 @@ SDValue DAGTypeLegalizer::PromoteIntRes_LOAD(LoadSDNode *N) {
   SDValue Res = DAG.getExtLoad(ExtType, dl, NVT, N->getChain(), N->getBasePtr(),
                                N->getMemoryVT(), N->getMemOperand());
 
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
   return Res;
@@ -489,20 +492,34 @@ SDValue DAGTypeLegalizer::PromoteIntRes_MLOAD(MaskedLoadSDNode *N) {
   EVT NVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
   SDValue ExtSrc0 = GetPromotedInteger(N->getSrc0());
 
-  SDValue Mask = N->getMask();
-  EVT NewMaskVT = getSetCCResultType(NVT);
-  if (NewMaskVT != N->getMask().getValueType())
-    Mask = PromoteTargetBoolean(Mask, NewMaskVT);
   SDLoc dl(N);
-
   SDValue Res = DAG.getMaskedLoad(NVT, dl, N->getChain(), N->getBasePtr(),
-                                  Mask, ExtSrc0, N->getMemoryVT(),
+                                  N->getMask(), ExtSrc0, N->getMemoryVT(),
                                   N->getMemOperand(), ISD::SEXTLOAD);
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
   return Res;
 }
+
+SDValue DAGTypeLegalizer::PromoteIntRes_MGATHER(MaskedGatherSDNode *N) {
+  EVT NVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
+  SDValue ExtSrc0 = GetPromotedInteger(N->getValue());
+  assert(NVT == ExtSrc0.getValueType() &&
+      "Gather result type and the passThru agrument type should be the same");
+
+  SDLoc dl(N);
+  SDValue Ops[] = {N->getChain(), ExtSrc0, N->getMask(), N->getBasePtr(),
+                   N->getIndex()};
+  SDValue Res = DAG.getMaskedGather(DAG.getVTList(NVT, MVT::Other),
+                                    N->getMemoryVT(), dl, Ops,
+                                    N->getMemOperand()); 
+  // Legalize the chain result - switch anything that used the old chain to
+  // use the new one.
+  ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
+  return Res;
+}
+
 /// Promote the overflow flag of an overflowing arithmetic node.
 SDValue DAGTypeLegalizer::PromoteIntRes_Overflow(SDNode *N) {
   // Simply change the return type of the boolean result.
@@ -889,6 +906,10 @@ bool DAGTypeLegalizer::PromoteIntegerOperand(SDNode *N, unsigned OpNo) {
                                                     OpNo); break;
   case ISD::MLOAD:        Res = PromoteIntOp_MLOAD(cast<MaskedLoadSDNode>(N),
                                                     OpNo); break;
+  case ISD::MGATHER:  Res = PromoteIntOp_MGATHER(cast<MaskedGatherSDNode>(N),
+                                                 OpNo); break;
+  case ISD::MSCATTER: Res = PromoteIntOp_MSCATTER(cast<MaskedScatterSDNode>(N),
+                                                  OpNo); break;
   case ISD::TRUNCATE:     Res = PromoteIntOp_TRUNCATE(N); break;
   case ISD::FP16_TO_FP:
   case ISD::UINT_TO_FP:   Res = PromoteIntOp_UINT_TO_FP(N); break;
@@ -1157,56 +1178,49 @@ SDValue DAGTypeLegalizer::PromoteIntOp_STORE(StoreSDNode *N, unsigned OpNo){
                            N->getMemoryVT(), N->getMemOperand());
 }
 
-SDValue DAGTypeLegalizer::PromoteIntOp_MSTORE(MaskedStoreSDNode *N, unsigned OpNo){
+SDValue DAGTypeLegalizer::PromoteIntOp_MSTORE(MaskedStoreSDNode *N,
+                                              unsigned OpNo) {
 
   SDValue DataOp = N->getValue();
   EVT DataVT = DataOp.getValueType();
   SDValue Mask = N->getMask();
-  EVT MaskVT = Mask.getValueType();
   SDLoc dl(N);
 
   bool TruncateStore = false;
-  if (!TLI.isTypeLegal(DataVT)) {
-    if (getTypeAction(DataVT) == TargetLowering::TypePromoteInteger) {
-      DataOp = GetPromotedInteger(DataOp);
-      if (!TLI.isTypeLegal(MaskVT))
-        Mask = PromoteTargetBoolean(Mask, DataOp.getValueType());
-      TruncateStore = true;
-    }
+  if (OpNo == 2) {
+    // Mask comes before the data operand. If the data operand is legal, we just
+    // promote the mask.
+    // When the data operand has illegal type, we should legalize the data
+    // operand first. The mask will be promoted/splitted/widened according to
+    // the data operand type.
+    if (TLI.isTypeLegal(DataVT))
+      Mask = PromoteTargetBoolean(Mask, DataVT);
     else {
-      assert(getTypeAction(DataVT) == TargetLowering::TypeWidenVector &&
-             "Unexpected data legalization in MSTORE");
-      DataOp = GetWidenedVector(DataOp);
-
-      if (getTypeAction(MaskVT) == TargetLowering::TypeWidenVector)
-        Mask = GetWidenedVector(Mask);
-      else {
-        EVT BoolVT = getSetCCResultType(DataOp.getValueType());
-
-        // We can't use ModifyToType() because we should fill the mask with
-        // zeroes
-        unsigned WidenNumElts = BoolVT.getVectorNumElements();
-        unsigned MaskNumElts = MaskVT.getVectorNumElements();
+      if (getTypeAction(DataVT) == TargetLowering::TypePromoteInteger)
+        return PromoteIntOp_MSTORE(N, 3);
 
-        unsigned NumConcat = WidenNumElts / MaskNumElts;
-        SmallVector<SDValue, 16> Ops(NumConcat);
-        SDValue ZeroVal = DAG.getConstant(0, dl, MaskVT);
-        Ops[0] = Mask;
-        for (unsigned i = 1; i != NumConcat; ++i)
-          Ops[i] = ZeroVal;
+      else if (getTypeAction(DataVT) == TargetLowering::TypeWidenVector)
+        return WidenVecOp_MSTORE(N, 3);
 
-        Mask = DAG.getNode(ISD::CONCAT_VECTORS, dl, BoolVT, Ops);
+      else {
+        assert (getTypeAction(DataVT) == TargetLowering::TypeSplitVector);
+        return SplitVecOp_MSTORE(N, 3);
       }
     }
+  } else { // Data operand
+    assert(OpNo == 3 && "Unexpected operand for promotion");
+    DataOp = GetPromotedInteger(DataOp);
+    Mask = PromoteTargetBoolean(Mask, DataOp.getValueType());
+    TruncateStore = true;
   }
-  else
-    Mask = PromoteTargetBoolean(N->getMask(), DataOp.getValueType());
+
   return DAG.getMaskedStore(N->getChain(), dl, DataOp, N->getBasePtr(), Mask,
                             N->getMemoryVT(), N->getMemOperand(),
                             TruncateStore);
 }
 
-SDValue DAGTypeLegalizer::PromoteIntOp_MLOAD(MaskedLoadSDNode *N, unsigned OpNo){
+SDValue DAGTypeLegalizer::PromoteIntOp_MLOAD(MaskedLoadSDNode *N,
+                                             unsigned OpNo) {
   assert(OpNo == 2 && "Only know how to promote the mask!");
   EVT DataVT = N->getValueType(0);
   SDValue Mask = PromoteTargetBoolean(N->getOperand(OpNo), DataVT);
@@ -1215,6 +1229,31 @@ SDValue DAGTypeLegalizer::PromoteIntOp_MLOAD(MaskedLoadSDNode *N, unsigned OpNo)
   return SDValue(DAG.UpdateNodeOperands(N, NewOps), 0);
 }
 
+SDValue DAGTypeLegalizer::PromoteIntOp_MGATHER(MaskedGatherSDNode *N,
+                                               unsigned OpNo) {
+
+  SmallVector<SDValue, 5> NewOps(N->op_begin(), N->op_end());
+  if (OpNo == 2) {
+    // The Mask
+    EVT DataVT = N->getValueType(0);
+    NewOps[OpNo] = PromoteTargetBoolean(N->getOperand(OpNo), DataVT);
+  } else
+    NewOps[OpNo] = GetPromotedInteger(N->getOperand(OpNo));
+  return SDValue(DAG.UpdateNodeOperands(N, NewOps), 0);
+}
+
+SDValue DAGTypeLegalizer::PromoteIntOp_MSCATTER(MaskedScatterSDNode *N,
+                                                unsigned OpNo) {
+  SmallVector<SDValue, 5> NewOps(N->op_begin(), N->op_end());
+  if (OpNo == 2) {
+    // The Mask
+    EVT DataVT = N->getValue().getValueType();
+    NewOps[OpNo] = PromoteTargetBoolean(N->getOperand(OpNo), DataVT);
+  } else
+    NewOps[OpNo] = GetPromotedInteger(N->getOperand(OpNo));
+  return SDValue(DAG.UpdateNodeOperands(N, NewOps), 0);
+}
+
 SDValue DAGTypeLegalizer::PromoteIntOp_TRUNCATE(SDNode *N) {
   SDValue Op = GetPromotedInteger(N->getOperand(0));
   return DAG.getNode(ISD::TRUNCATE, SDLoc(N), N->getValueType(0), Op);
@@ -2071,7 +2110,7 @@ void DAGTypeLegalizer::ExpandIntRes_LOAD(LoadSDNode *N,
     }
   }
 
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(N, 1), Ch);
 }

diff --git a/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp b/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp
index d6b4f7921f2babbf242e347a5a001a41a5b2846e..2a0b0aa4479488fe3a3a6d25cefcfc77bd3c29de 100644 (file)
--- a/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp
+++ b/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp
@@ -1127,6 +1127,23 @@ SDValue DAGTypeLegalizer::PromoteTargetBoolean(SDValue Bool, EVT ValVT) {
   return DAG.getNode(ExtendCode, dl, BoolVT, Bool);
 }
 
+/// WidenTargetBoolean - Widen the given target boolean to a target boolean
+/// of the given type. The boolean vector is widened and then promoted to match
+/// the target boolean type of the given ValVT.
+SDValue DAGTypeLegalizer::WidenTargetBoolean(SDValue Bool, EVT ValVT,
+                                             bool WithZeroes) {
+  SDLoc dl(Bool);
+  EVT BoolVT = Bool.getValueType();
+
+  assert(ValVT.getVectorNumElements() > BoolVT.getVectorNumElements() &&
+         TLI.isTypeLegal(ValVT) &&
+         "Unexpected types in WidenTargetBoolean");
+  EVT WideVT = EVT::getVectorVT(*DAG.getContext(), BoolVT.getScalarType(),
+                                ValVT.getVectorNumElements());
+  Bool = ModifyToType(Bool, WideVT, WithZeroes);
+  return PromoteTargetBoolean(Bool, ValVT);
+}
+
 /// SplitInteger - Return the lower LoVT bits of Op in Lo and the upper HiVT
 /// bits in Hi.
 void DAGTypeLegalizer::SplitInteger(SDValue Op,

diff --git a/lib/CodeGen/SelectionDAG/LegalizeTypes.h b/lib/CodeGen/SelectionDAG/LegalizeTypes.h
index 84ea374345e990d384996357cd5226cd9c27a400..7994f4dad4a9a4e459e8270d868cda347c253660 100644 (file)
--- a/lib/CodeGen/SelectionDAG/LegalizeTypes.h
+++ b/lib/CodeGen/SelectionDAG/LegalizeTypes.h
@@ -187,6 +187,11 @@ private:
   std::pair<SDValue, SDValue> ExpandAtomic(SDNode *Node);
 
   SDValue PromoteTargetBoolean(SDValue Bool, EVT ValVT);
+
+  /// Modify Bit Vector to match SetCC result type of ValVT.
+  /// The bit vector is widened with zeroes when WithZeroes is true.
+  SDValue WidenTargetBoolean(SDValue Bool, EVT ValVT, bool WithZeroes = false);
+
   void ReplaceValueWith(SDValue From, SDValue To);
   void SplitInteger(SDValue Op, SDValue &Lo, SDValue &Hi);
   void SplitInteger(SDValue Op, EVT LoVT, EVT HiVT,
@@ -261,6 +266,7 @@ private:
   SDValue PromoteIntRes_INT_EXTEND(SDNode *N);
   SDValue PromoteIntRes_LOAD(LoadSDNode *N);
   SDValue PromoteIntRes_MLOAD(MaskedLoadSDNode *N);
+  SDValue PromoteIntRes_MGATHER(MaskedGatherSDNode *N);
   SDValue PromoteIntRes_Overflow(SDNode *N);
   SDValue PromoteIntRes_SADDSUBO(SDNode *N, unsigned ResNo);
   SDValue PromoteIntRes_SDIV(SDNode *N);
@@ -307,6 +313,8 @@ private:
   SDValue PromoteIntOp_ZERO_EXTEND(SDNode *N);
   SDValue PromoteIntOp_MSTORE(MaskedStoreSDNode *N, unsigned OpNo);
   SDValue PromoteIntOp_MLOAD(MaskedLoadSDNode *N, unsigned OpNo);
+  SDValue PromoteIntOp_MSCATTER(MaskedScatterSDNode *N, unsigned OpNo);
+  SDValue PromoteIntOp_MGATHER(MaskedGatherSDNode *N, unsigned OpNo);
 
   void PromoteSetCCOperands(SDValue &LHS,SDValue &RHS, ISD::CondCode Code);
 
@@ -710,6 +718,7 @@ private:
   SDValue WidenVecRes_INSERT_VECTOR_ELT(SDNode* N);
   SDValue WidenVecRes_LOAD(SDNode* N);
   SDValue WidenVecRes_MLOAD(MaskedLoadSDNode* N);
+  SDValue WidenVecRes_MGATHER(MaskedGatherSDNode* N);
   SDValue WidenVecRes_SCALAR_TO_VECTOR(SDNode* N);
   SDValue WidenVecRes_SELECT(SDNode* N);
   SDValue WidenVecRes_SELECT_CC(SDNode* N);
@@ -737,6 +746,7 @@ private:
   SDValue WidenVecOp_EXTRACT_SUBVECTOR(SDNode *N);
   SDValue WidenVecOp_STORE(SDNode* N);
   SDValue WidenVecOp_MSTORE(SDNode* N, unsigned OpNo);
+  SDValue WidenVecOp_MSCATTER(SDNode* N, unsigned OpNo);
   SDValue WidenVecOp_SETCC(SDNode* N);
 
   SDValue WidenVecOp_Convert(SDNode *N);
@@ -776,8 +786,10 @@ private:
 
   /// Modifies a vector input (widen or narrows) to a vector of NVT.  The
   /// input vector must have the same element type as NVT.
-  SDValue ModifyToType(SDValue InOp, EVT WidenVT);
-
+  /// When FillWithZeroes is "on" the vector will be widened with
+  /// zeroes.
+  /// By default, the vector will be widened with undefined values.
+  SDValue ModifyToType(SDValue InOp, EVT NVT, bool FillWithZeroes = false);
 
   //===--------------------------------------------------------------------===//
   // Generic Splitting: LegalizeTypesGeneric.cpp

diff --git a/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp b/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
index 5e4d469df37fe2b0ca46c3f21365debdc097f2b4..d0187d36dee21f9e05ee647df58a4a5c6958748b 100644 (file)
--- a/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
+++ b/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
@@ -235,7 +235,7 @@ SDValue DAGTypeLegalizer::ScalarizeVecRes_LOAD(LoadSDNode *N) {
                                N->isInvariant(), N->getOriginalAlignment(),
                                N->getAAInfo());
 
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(N, 1), Result.getValue(1));
   return Result;
@@ -1020,7 +1020,7 @@ void DAGTypeLegalizer::SplitVecRes_LOAD(LoadSDNode *LD, SDValue &Lo,
   Ch = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Lo.getValue(1),
                    Hi.getValue(1));
 
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(LD, 1), Ch);
 }
@@ -1034,6 +1034,7 @@ void DAGTypeLegalizer::SplitVecRes_MLOAD(MaskedLoadSDNode *MLD,
   SDValue Ch = MLD->getChain();
   SDValue Ptr = MLD->getBasePtr();
   SDValue Mask = MLD->getMask();
+  SDValue Src0 = MLD->getSrc0();
   unsigned Alignment = MLD->getOriginalAlignment();
   ISD::LoadExtType ExtType = MLD->getExtensionType();
 
@@ -1043,16 +1044,22 @@ void DAGTypeLegalizer::SplitVecRes_MLOAD(MaskedLoadSDNode *MLD,
     (Alignment == MLD->getValueType(0).getSizeInBits()/8) ?
      Alignment/2 : Alignment;
 
+  // Split Mask operand
   SDValue MaskLo, MaskHi;
-  std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, dl);
+  if (getTypeAction(Mask.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Mask, MaskLo, MaskHi);
+  else
+    std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, dl);
 
   EVT MemoryVT = MLD->getMemoryVT();
   EVT LoMemVT, HiMemVT;
   std::tie(LoMemVT, HiMemVT) = DAG.GetSplitDestVTs(MemoryVT);
 
-  SDValue Src0 = MLD->getSrc0();
   SDValue Src0Lo, Src0Hi;
-  std::tie(Src0Lo, Src0Hi) = DAG.SplitVector(Src0, dl);
+  if (getTypeAction(Src0.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Src0, Src0Lo, Src0Hi);
+  else
+    std::tie(Src0Lo, Src0Hi) = DAG.SplitVector(Src0, dl);
 
   MachineMemOperand *MMO = DAG.getMachineFunction().
     getMachineMemOperand(MLD->getPointerInfo(), 
@@ -1080,7 +1087,7 @@ void DAGTypeLegalizer::SplitVecRes_MLOAD(MaskedLoadSDNode *MLD,
   Ch = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Lo.getValue(1),
                    Hi.getValue(1));
 
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(MLD, 1), Ch);
 
@@ -1095,20 +1102,33 @@ void DAGTypeLegalizer::SplitVecRes_MGATHER(MaskedGatherSDNode *MGT,
   SDValue Ch = MGT->getChain();
   SDValue Ptr = MGT->getBasePtr();
   SDValue Mask = MGT->getMask();
+  SDValue Src0 = MGT->getValue();
+  SDValue Index = MGT->getIndex();
   unsigned Alignment = MGT->getOriginalAlignment();
 
+  // Split Mask operand
   SDValue MaskLo, MaskHi;
-  std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, dl);
+  if (getTypeAction(Mask.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Mask, MaskLo, MaskHi);
+  else
+    std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, dl);
 
   EVT MemoryVT = MGT->getMemoryVT();
   EVT LoMemVT, HiMemVT;
+  // Split MemoryVT
   std::tie(LoMemVT, HiMemVT) = DAG.GetSplitDestVTs(MemoryVT);
 
   SDValue Src0Lo, Src0Hi;
-  std::tie(Src0Lo, Src0Hi) = DAG.SplitVector(MGT->getValue(), dl);
+  if (getTypeAction(Src0.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Src0, Src0Lo, Src0Hi);
+  else
+    std::tie(Src0Lo, Src0Hi) = DAG.SplitVector(Src0, dl);
 
   SDValue IndexHi, IndexLo;
-  std::tie(IndexLo, IndexHi) = DAG.SplitVector(MGT->getIndex(), dl);
+  if (getTypeAction(Index.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Index, IndexLo, IndexHi);
+  else
+    std::tie(IndexLo, IndexHi) = DAG.SplitVector(Index, dl);
 
   MachineMemOperand *MMO = DAG.getMachineFunction().
     getMachineMemOperand(MGT->getPointerInfo(), 
@@ -1128,7 +1148,7 @@ void DAGTypeLegalizer::SplitVecRes_MGATHER(MaskedGatherSDNode *MGT,
   Ch = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Lo.getValue(1),
                    Hi.getValue(1));
 
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(MGT, 1), Ch);
 }
@@ -1599,23 +1619,31 @@ SDValue DAGTypeLegalizer::SplitVecOp_MGATHER(MaskedGatherSDNode *MGT,
   SDValue Ptr = MGT->getBasePtr();
   SDValue Index = MGT->getIndex();
   SDValue Mask = MGT->getMask();
+  SDValue Src0 = MGT->getValue();
   unsigned Alignment = MGT->getOriginalAlignment();
 
   SDValue MaskLo, MaskHi;
-  std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, dl);
+  if (getTypeAction(Mask.getValueType()) == TargetLowering::TypeSplitVector)
+    // Split Mask operand
+    GetSplitVector(Mask, MaskLo, MaskHi);
+  else
+    std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, dl);
 
   EVT MemoryVT = MGT->getMemoryVT();
   EVT LoMemVT, HiMemVT;
   std::tie(LoMemVT, HiMemVT) = DAG.GetSplitDestVTs(MemoryVT);
 
   SDValue Src0Lo, Src0Hi;
-  std::tie(Src0Lo, Src0Hi) = DAG.SplitVector(MGT->getValue(), dl);
+  if (getTypeAction(Src0.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Src0, Src0Lo, Src0Hi);
+  else
+    std::tie(Src0Lo, Src0Hi) = DAG.SplitVector(Src0, dl);
 
   SDValue IndexHi, IndexLo;
-  if (Index.getNode())
-    std::tie(IndexLo, IndexHi) = DAG.SplitVector(Index, dl);
+  if (getTypeAction(Index.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Index, IndexLo, IndexHi);
   else
-    IndexLo = IndexHi = Index;
+    std::tie(IndexLo, IndexHi) = DAG.SplitVector(Index, dl);
 
   MachineMemOperand *MMO = DAG.getMachineFunction().
     getMachineMemOperand(MGT->getPointerInfo(), 
@@ -1641,7 +1669,7 @@ SDValue DAGTypeLegalizer::SplitVecOp_MGATHER(MaskedGatherSDNode *MGT,
   Ch = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Lo.getValue(1),
                    Hi.getValue(1));
 
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(MGT, 1), Ch);
 
@@ -1665,9 +1693,21 @@ SDValue DAGTypeLegalizer::SplitVecOp_MSTORE(MaskedStoreSDNode *N,
   std::tie(LoMemVT, HiMemVT) = DAG.GetSplitDestVTs(MemoryVT);
 
   SDValue DataLo, DataHi;
-  GetSplitVector(Data, DataLo, DataHi);
+  if (getTypeAction(Data.getValueType()) == TargetLowering::TypeSplitVector)
+    // Split Data operand
+    GetSplitVector(Data, DataLo, DataHi);
+  else
+    std::tie(DataLo, DataHi) = DAG.SplitVector(Data, DL);
+
   SDValue MaskLo, MaskHi;
-  GetSplitVector(Mask, MaskLo, MaskHi);
+  if (getTypeAction(Mask.getValueType()) == TargetLowering::TypeSplitVector)
+    // Split Mask operand
+    GetSplitVector(Mask, MaskLo, MaskHi);
+  else
+    std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, DL);
+
+  MaskLo = PromoteTargetBoolean(MaskLo, DataLo.getValueType());
+  MaskHi = PromoteTargetBoolean(MaskHi, DataHi.getValueType());
 
   // if Alignment is equal to the vector size,
   // take the half of it for the second part
@@ -1712,25 +1752,29 @@ SDValue DAGTypeLegalizer::SplitVecOp_MSCATTER(MaskedScatterSDNode *N,
   unsigned Alignment = N->getOriginalAlignment();
   SDLoc DL(N);
 
+  // Split all operands
   EVT LoMemVT, HiMemVT;
   std::tie(LoMemVT, HiMemVT) = DAG.GetSplitDestVTs(MemoryVT);
 
   SDValue DataLo, DataHi;
-  GetSplitVector(Data, DataLo, DataHi);
-  SDValue MaskLo, MaskHi;
-  GetSplitVector(Mask, MaskLo, MaskHi);
+  if (getTypeAction(Data.getValueType()) == TargetLowering::TypeSplitVector)
+    // Split Data operand
+    GetSplitVector(Data, DataLo, DataHi);
+  else
+    std::tie(DataLo, DataHi) = DAG.SplitVector(Data, DL);
 
-    SDValue PtrLo, PtrHi;
-  if (Ptr.getValueType().isVector()) // gather form vector of pointers
-    std::tie(PtrLo, PtrHi) = DAG.SplitVector(Ptr, DL);
+  SDValue MaskLo, MaskHi;
+  if (getTypeAction(Mask.getValueType()) == TargetLowering::TypeSplitVector)
+    // Split Mask operand
+    GetSplitVector(Mask, MaskLo, MaskHi);
   else
-    PtrLo = PtrHi = Ptr;
+    std::tie(MaskLo, MaskHi) = DAG.SplitVector(Mask, DL);
 
   SDValue IndexHi, IndexLo;
-  if (Index.getNode())
-    std::tie(IndexLo, IndexHi) = DAG.SplitVector(Index, DL);
+  if (getTypeAction(Index.getValueType()) == TargetLowering::TypeSplitVector)
+    GetSplitVector(Index, IndexLo, IndexHi);
   else
-    IndexLo = IndexHi = Index;
+    std::tie(IndexLo, IndexHi) = DAG.SplitVector(Index, DL);
 
   SDValue Lo, Hi;
   MachineMemOperand *MMO = DAG.getMachineFunction().
@@ -1738,7 +1782,7 @@ SDValue DAGTypeLegalizer::SplitVecOp_MSCATTER(MaskedScatterSDNode *N,
                          MachineMemOperand::MOStore, LoMemVT.getStoreSize(),
                          Alignment, N->getAAInfo(), N->getRanges());
 
-  SDValue OpsLo[] = {Ch, DataLo, MaskLo, PtrLo, IndexLo};
+  SDValue OpsLo[] = {Ch, DataLo, MaskLo, Ptr, IndexLo};
   Lo = DAG.getMaskedScatter(DAG.getVTList(MVT::Other), DataLo.getValueType(),
                             DL, OpsLo, MMO);
 
@@ -1747,7 +1791,7 @@ SDValue DAGTypeLegalizer::SplitVecOp_MSCATTER(MaskedScatterSDNode *N,
                          MachineMemOperand::MOStore,  HiMemVT.getStoreSize(),
                          Alignment, N->getAAInfo(), N->getRanges());
 
-  SDValue OpsHi[] = {Ch, DataHi, MaskHi, PtrHi, IndexHi};
+  SDValue OpsHi[] = {Ch, DataHi, MaskHi, Ptr, IndexHi};
   Hi = DAG.getMaskedScatter(DAG.getVTList(MVT::Other), DataHi.getValueType(),
                             DL, OpsHi, MMO);
 
@@ -1975,6 +2019,9 @@ void DAGTypeLegalizer::WidenVectorResult(SDNode *N, unsigned ResNo) {
   case ISD::MLOAD:
     Res = WidenVecRes_MLOAD(cast<MaskedLoadSDNode>(N));
     break;
+  case ISD::MGATHER:
+    Res = WidenVecRes_MGATHER(cast<MaskedGatherSDNode>(N));
+    break;
 
   case ISD::ADD:
   case ISD::AND:
@@ -2728,7 +2775,35 @@ SDValue DAGTypeLegalizer::WidenVecRes_MLOAD(MaskedLoadSDNode *N) {
   SDValue Res = DAG.getMaskedLoad(WidenVT, dl, N->getChain(), N->getBasePtr(),
                                   Mask, Src0, N->getMemoryVT(),
                                   N->getMemOperand(), ExtType);
-  // Legalized the chain result - switch anything that used the old chain to
+  // Legalize the chain result - switch anything that used the old chain to
+  // use the new one.
+  ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
+  return Res;
+}
+
+SDValue DAGTypeLegalizer::WidenVecRes_MGATHER(MaskedGatherSDNode *N) {
+
+  EVT WideVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
+  SDValue Mask = N->getMask();
+  SDValue Src0 = GetWidenedVector(N->getValue());
+  unsigned NumElts = WideVT.getVectorNumElements();
+  SDLoc dl(N);
+
+  // The mask should be widened as well
+  Mask = WidenTargetBoolean(Mask, WideVT, true);
+
+  // Widen the Index operand
+  SDValue Index = N->getIndex();
+  EVT WideIndexVT = EVT::getVectorVT(*DAG.getContext(),
+                                     Index.getValueType().getScalarType(),
+                                     NumElts);
+  Index = ModifyToType(Index, WideIndexVT);
+  SDValue Ops[] = { N->getChain(), Src0, Mask, N->getBasePtr(), Index };
+  SDValue Res = DAG.getMaskedGather(DAG.getVTList(WideVT, MVT::Other),
+                                    N->getMemoryVT(), dl, Ops,
+                                    N->getMemOperand());
+
+  // Legalize the chain result - switch anything that used the old chain to
   // use the new one.
   ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
   return Res;
@@ -2890,6 +2965,7 @@ bool DAGTypeLegalizer::WidenVectorOperand(SDNode *N, unsigned OpNo) {
   case ISD::EXTRACT_VECTOR_ELT: Res = WidenVecOp_EXTRACT_VECTOR_ELT(N); break;
   case ISD::STORE:              Res = WidenVecOp_STORE(N); break;
   case ISD::MSTORE:             Res = WidenVecOp_MSTORE(N, OpNo); break;
+  case ISD::MSCATTER:           Res = WidenVecOp_MSCATTER(N, OpNo); break;
   case ISD::SETCC:              Res = WidenVecOp_SETCC(N); break;
   case ISD::FCOPYSIGN:          Res = WidenVecOp_FCOPYSIGN(N); break;
 
@@ -3137,6 +3213,34 @@ SDValue DAGTypeLegalizer::WidenVecOp_MSTORE(SDNode *N, unsigned OpNo) {
                             false);
 }
 
+SDValue DAGTypeLegalizer::WidenVecOp_MSCATTER(SDNode *N, unsigned OpNo) {
+  assert(OpNo == 1 && "Can widen only data operand of mscatter");
+  MaskedScatterSDNode *MSC = cast<MaskedScatterSDNode>(N);
+  SDValue DataOp = MSC->getValue();
+  SDValue Mask = MSC->getMask();
+
+  // Widen the value
+  SDValue WideVal = GetWidenedVector(DataOp);
+  EVT WideVT = WideVal.getValueType();
+  unsigned NumElts = WideVal.getValueType().getVectorNumElements();
+  SDLoc dl(N);
+
+  // The mask should be widened as well
+  Mask = WidenTargetBoolean(Mask, WideVT, true);
+
+  // Widen index
+  SDValue Index = MSC->getIndex();
+  EVT WideIndexVT = EVT::getVectorVT(*DAG.getContext(),
+                                     Index.getValueType().getScalarType(),
+                                     NumElts);
+  Index = ModifyToType(Index, WideIndexVT);
+
+  SDValue Ops[] = {MSC->getChain(), WideVal, Mask, MSC->getBasePtr(), Index};
+  return DAG.getMaskedScatter(DAG.getVTList(MVT::Other),
+                              MSC->getMemoryVT(), dl, Ops,
+                              MSC->getMemOperand());
+}
+
 SDValue DAGTypeLegalizer::WidenVecOp_SETCC(SDNode *N) {
   SDValue InOp0 = GetWidenedVector(N->getOperand(0));
   SDValue InOp1 = GetWidenedVector(N->getOperand(1));
@@ -3600,7 +3704,9 @@ DAGTypeLegalizer::GenWidenVectorTruncStores(SmallVectorImpl<SDValue> &StChain,
 
 /// Modifies a vector input (widen or narrows) to a vector of NVT.  The
 /// input vector must have the same element type as NVT.
-SDValue DAGTypeLegalizer::ModifyToType(SDValue InOp, EVT NVT) {
+/// FillWithZeroes specifies that the vector should be widened with zeroes.
+SDValue DAGTypeLegalizer::ModifyToType(SDValue InOp, EVT NVT,
+                                       bool FillWithZeroes) {
   // Note that InOp might have been widened so it might already have
   // the right width or it might need be narrowed.
   EVT InVT = InOp.getValueType();
@@ -3617,10 +3723,11 @@ SDValue DAGTypeLegalizer::ModifyToType(SDValue InOp, EVT NVT) {
   if (WidenNumElts > InNumElts && WidenNumElts % InNumElts == 0) {
     unsigned NumConcat = WidenNumElts / InNumElts;
     SmallVector<SDValue, 16> Ops(NumConcat);
-    SDValue UndefVal = DAG.getUNDEF(InVT);
+    SDValue FillVal = FillWithZeroes ? DAG.getConstant(0, dl, InVT) :
+      DAG.getUNDEF(InVT);
     Ops[0] = InOp;
     for (unsigned i = 1; i != NumConcat; ++i)
-      Ops[i] = UndefVal;
+      Ops[i] = FillVal;
 
     return DAG.getNode(ISD::CONCAT_VECTORS, dl, NVT, Ops);
   }
@@ -3640,8 +3747,9 @@ SDValue DAGTypeLegalizer::ModifyToType(SDValue InOp, EVT NVT) {
         ISD::EXTRACT_VECTOR_ELT, dl, EltVT, InOp,
         DAG.getConstant(Idx, dl, TLI.getVectorIdxTy(DAG.getDataLayout())));
 
-  SDValue UndefVal = DAG.getUNDEF(EltVT);
+  SDValue FillVal = FillWithZeroes ? DAG.getConstant(0, dl, EltVT) :
+    DAG.getUNDEF(EltVT);
   for ( ; Idx < WidenNumElts; ++Idx)
-    Ops[Idx] = UndefVal;
+    Ops[Idx] = FillVal;
   return DAG.getNode(ISD::BUILD_VECTOR, dl, NVT, Ops);
 }

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 829d6302bcee65eeecbac5bcc595c27f9a8c8a29..3765d64617bc90f03863516093c2c1bba6013382 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -1579,7 +1579,7 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
         setOperationAction(ISD::OR,  VT, Legal);
         setOperationAction(ISD::XOR,  VT, Legal);
       }
-      if (EltSize >= 32 && VT.getSizeInBits() <= 512) {
+      if ((VT.is128BitVector() || VT.is256BitVector()) && EltSize >= 32) {
         setOperationAction(ISD::MGATHER,  VT, Custom);
         setOperationAction(ISD::MSCATTER, VT, Custom);
       }
@@ -1605,6 +1605,8 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
         setOperationAction(ISD::INSERT_SUBVECTOR,    VT, Custom);
         setOperationAction(ISD::MLOAD,               VT, Legal);
         setOperationAction(ISD::MSTORE,              VT, Legal);
+        setOperationAction(ISD::MGATHER,  VT, Legal);
+        setOperationAction(ISD::MSCATTER, VT, Custom);
       }
     }
     for (auto VT : { MVT::v64i8, MVT::v32i16, MVT::v16i32 }) {
@@ -1813,6 +1815,8 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
   setTargetDAGCombine(ISD::BUILD_VECTOR);
   setTargetDAGCombine(ISD::MUL);
   setTargetDAGCombine(ISD::XOR);
+  setTargetDAGCombine(ISD::MSCATTER);
+  setTargetDAGCombine(ISD::MGATHER);
 
   computeRegisterProperties(Subtarget->getRegisterInfo());
 
@@ -19760,6 +19764,16 @@ static SDValue ExtendToType(SDValue InOp, MVT NVT, SelectionDAG &DAG,
   EVT EltVT = NVT.getVectorElementType();
 
   SDLoc dl(InOp);
+  if (InOp.getOpcode() == ISD::CONCAT_VECTORS &&
+      InOp.getNumOperands() == 2) {
+    SDValue N1 = InOp.getOperand(1);
+    if ((ISD::isBuildVectorAllZeros(N1.getNode()) && FillWithZeroes) ||
+        N1.isUndef()) {
+      InOp = InOp.getOperand(0);
+      InVT = InOp.getSimpleValueType();
+      InNumElts = InVT.getVectorNumElements();
+    }
+  }
   if (ISD::isBuildVectorOfConstantSDNodes(InOp.getNode()) ||
       ISD::isBuildVectorOfConstantFPSDNodes(InOp.getNode())) {
     SmallVector<SDValue, 16> Ops;
@@ -19783,28 +19797,93 @@ static SDValue LowerMSCATTER(SDValue Op, const X86Subtarget *Subtarget,
   assert(Subtarget->hasAVX512() &&
          "MGATHER/MSCATTER are supported on AVX-512 arch only");
 
+  // X86 scatter kills mask register, so its type should be added to
+  // the list of return values.
+  // If the "scatter" has 2 return values, it is already handled.
+  if (Op.getNode()->getNumValues() == 2)
+    return Op;
+
   MaskedScatterSDNode *N = cast<MaskedScatterSDNode>(Op.getNode());
-  MVT VT = N->getValue().getSimpleValueType();
+  SDValue Src = N->getValue();
+  MVT VT = Src.getSimpleValueType();
   assert(VT.getScalarSizeInBits() >= 32 && "Unsupported scatter op");
   SDLoc dl(Op);
 
-  // X86 scatter kills mask register, so its type should be added to
-  // the list of return values
-  if (N->getNumValues() == 1) {
-    SDValue Index = N->getIndex();
-    if (!Subtarget->hasVLX() && !VT.is512BitVector() &&
-        !Index.getSimpleValueType().is512BitVector())
-      Index = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::v8i64, Index);
-
-    SDVTList VTs = DAG.getVTList(N->getMask().getValueType(), MVT::Other);
-    SDValue Ops[] = { N->getOperand(0), N->getOperand(1),  N->getOperand(2),
-                      N->getOperand(3), Index };
-
-    SDValue NewScatter = DAG.getMaskedScatter(VTs, VT, dl, Ops, N->getMemOperand());
-    DAG.ReplaceAllUsesWith(Op, SDValue(NewScatter.getNode(), 1));
-    return SDValue(NewScatter.getNode(), 0);
+  SDValue NewScatter;
+  SDValue Index = N->getIndex();
+  SDValue Mask = N->getMask();
+  SDValue Chain = N->getChain();
+  SDValue BasePtr = N->getBasePtr();
+  MVT MemVT = N->getMemoryVT().getSimpleVT();
+  MVT IndexVT = Index.getSimpleValueType();
+  MVT MaskVT = Mask.getSimpleValueType();
+
+  if (MemVT.getScalarSizeInBits() < VT.getScalarSizeInBits()) {
+    // The v2i32 value was promoted to v2i64.
+    // Now we "redo" the type legalizer's work and widen the original
+    // v2i32 value to v4i32. The original v2i32 is retrieved from v2i64
+    // with a shuffle.
+    assert((MemVT == MVT::v2i32 && VT == MVT::v2i64) &&
+           "Unexpected memory type");
+    int ShuffleMask[] = {0, 2, -1, -1};
+    Src = DAG.getVectorShuffle(MVT::v4i32, dl, DAG.getBitcast(MVT::v4i32, Src),
+                               DAG.getUNDEF(MVT::v4i32), ShuffleMask);
+    // Now we have 4 elements instead of 2.
+    // Expand the index.
+    MVT NewIndexVT = MVT::getVectorVT(IndexVT.getScalarType(), 4);
+    Index = ExtendToType(Index, NewIndexVT, DAG);
+
+    // Expand the mask with zeroes
+    // Mask may be <2 x i64> or <2 x i1> at this moment
+    assert((MaskVT == MVT::v2i1 || MaskVT == MVT::v2i64) &&
+           "Unexpected mask type");
+    MVT ExtMaskVT = MVT::getVectorVT(MaskVT.getScalarType(), 4);
+    Mask = ExtendToType(Mask, ExtMaskVT, DAG, true);
+    VT = MVT::v4i32;
   }
-  return Op;
+
+  unsigned NumElts = VT.getVectorNumElements();
+  if (!Subtarget->hasVLX() && !VT.is512BitVector() &&
+      !Index.getSimpleValueType().is512BitVector()) {
+    // AVX512F supports only 512-bit vectors. Or data or index should
+    // be 512 bit wide. If now the both index and data are 256-bit, but
+    // the vector contains 8 elements, we just sign-extend the index
+    if (IndexVT == MVT::v8i32)
+      // Just extend index
+      Index = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::v8i64, Index);
+    else {
+      // The minimal number of elts in scatter is 8
+      NumElts = 8;
+      // Index
+      MVT NewIndexVT = MVT::getVectorVT(IndexVT.getScalarType(), NumElts);
+      // Use original index here, do not modify the index twice
+      Index = ExtendToType(N->getIndex(), NewIndexVT, DAG);
+      if (IndexVT.getScalarType() == MVT::i32)
+        Index = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::v8i64, Index);
+
+      // Mask
+      // At this point we have promoted mask operand
+      assert(MaskVT.getScalarSizeInBits() >= 32 && "unexpected mask type");
+      MVT ExtMaskVT = MVT::getVectorVT(MaskVT.getScalarType(), NumElts);
+      // Use the original mask here, do not modify the mask twice
+      Mask = ExtendToType(N->getMask(), ExtMaskVT, DAG, true); 
+
+      // The value that should be stored
+      MVT NewVT = MVT::getVectorVT(VT.getScalarType(), NumElts);
+      Src = ExtendToType(Src, NewVT, DAG);
+    }
+  }
+  // If the mask is "wide" at this point - truncate it to i1 vector
+  MVT BitMaskVT = MVT::getVectorVT(MVT::i1, NumElts);
+  Mask = DAG.getNode(ISD::TRUNCATE, dl, BitMaskVT, Mask);
+
+  // The mask is killed by scatter, add it to the values
+  SDVTList VTs = DAG.getVTList(BitMaskVT, MVT::Other);
+  SDValue Ops[] = {Chain, Src, Mask, BasePtr, Index};
+  NewScatter = DAG.getMaskedScatter(VTs, N->getMemoryVT(), dl, Ops,
+                                    N->getMemOperand());
+  DAG.ReplaceAllUsesWith(Op, SDValue(NewScatter.getNode(), 1));
+  return SDValue(NewScatter.getNode(), 0);
 }
 
 static SDValue LowerMLOAD(SDValue Op, const X86Subtarget *Subtarget,
@@ -19869,17 +19948,59 @@ static SDValue LowerMGATHER(SDValue Op, const X86Subtarget *Subtarget,
          "MGATHER/MSCATTER are supported on AVX-512 arch only");
 
   MaskedGatherSDNode *N = cast<MaskedGatherSDNode>(Op.getNode());
+  SDLoc dl(Op);
   MVT VT = Op.getSimpleValueType();
+  SDValue Index = N->getIndex();
+  SDValue Mask = N->getMask();
+  SDValue Src0 = N->getValue();
+  MVT IndexVT = Index.getSimpleValueType();
+  MVT MaskVT = Mask.getSimpleValueType();
+
+  unsigned NumElts = VT.getVectorNumElements();
   assert(VT.getScalarSizeInBits() >= 32 && "Unsupported gather op");
-  SDLoc dl(Op);
 
-  SDValue Index = N->getIndex();
   if (!Subtarget->hasVLX() && !VT.is512BitVector() &&
       !Index.getSimpleValueType().is512BitVector()) {
-    Index = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::v8i64, Index);
-    SDValue Ops[] = { N->getOperand(0), N->getOperand(1),  N->getOperand(2),
-                      N->getOperand(3), Index };
-    DAG.UpdateNodeOperands(N, Ops);
+    // AVX512F supports only 512-bit vectors. Or data or index should
+    // be 512 bit wide. If now the both index and data are 256-bit, but
+    // the vector contains 8 elements, we just sign-extend the index
+    if (NumElts == 8) {
+      Index = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::v8i64, Index);
+      SDValue Ops[] = { N->getOperand(0), N->getOperand(1),  N->getOperand(2),
+                        N->getOperand(3), Index };
+      DAG.UpdateNodeOperands(N, Ops);
+      return Op;
+    }
+
+    // Minimal number of elements in Gather
+    NumElts = 8;
+    // Index
+    MVT NewIndexVT = MVT::getVectorVT(IndexVT.getScalarType(), NumElts);
+    Index = ExtendToType(Index, NewIndexVT, DAG);
+    if (IndexVT.getScalarType() == MVT::i32)
+      Index = DAG.getNode(ISD::SIGN_EXTEND, dl, MVT::v8i64, Index);
+
+    // Mask
+    MVT MaskBitVT = MVT::getVectorVT(MVT::i1, NumElts);
+    // At this point we have promoted mask operand
+    assert(MaskVT.getScalarSizeInBits() >= 32 && "unexpected mask type");
+    MVT ExtMaskVT = MVT::getVectorVT(MaskVT.getScalarType(), NumElts);
+    Mask = ExtendToType(Mask, ExtMaskVT, DAG, true);
+    Mask = DAG.getNode(ISD::TRUNCATE, dl, MaskBitVT, Mask);
+
+    // The pass-thru value
+    MVT NewVT = MVT::getVectorVT(VT.getScalarType(), NumElts);
+    Src0 = ExtendToType(Src0, NewVT, DAG);
+
+    SDValue Ops[] = { N->getChain(), Src0, Mask, N->getBasePtr(), Index };
+    SDValue NewGather = DAG.getMaskedGather(DAG.getVTList(NewVT, MVT::Other),
+                                            N->getMemoryVT(), dl, Ops,
+                                            N->getMemOperand());
+    SDValue Exract = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, VT,
+                                 NewGather.getValue(0),
+                                 DAG.getIntPtrConstant(0, dl));
+    SDValue RetOps[] = {Exract, NewGather.getValue(1)};
+    return DAG.getMergeValues(RetOps, dl);
   }
   return Op;
 }
@@ -26907,6 +27028,20 @@ static SDValue PerformBLENDICombine(SDNode *N, SelectionDAG &DAG) {
   return SDValue();
 }
 
+static SDValue PerformGatherScatterCombine(SDNode *N, SelectionDAG &DAG) {
+  SDLoc DL(N);
+  // Gather and Scatter instructions use k-registers for masks. The type of
+  // the masks is v*i1. So the mask will be truncated anyway.
+  // The SIGN_EXTEND_INREG my be dropped.
+  SDValue Mask = N->getOperand(2);
+  if (Mask.getOpcode() == ISD::SIGN_EXTEND_INREG) {
+    SmallVector<SDValue, 5> NewOps(N->op_begin(), N->op_end());
+    NewOps[2] = Mask.getOperand(0);
+    DAG.UpdateNodeOperands(N, NewOps);
+  }
+  return SDValue();
+}
+
 // Helper function of PerformSETCCCombine. It is to materialize "setb reg"
 // as "sbb reg,reg", since it can be extended without zext and produces
 // an all-ones bit which is more useful than 0/1 in some cases.
@@ -27348,6 +27483,8 @@ SDValue X86TargetLowering::PerformDAGCombine(SDNode *N,
   case ISD::VECTOR_SHUFFLE: return PerformShuffleCombine(N, DAG, DCI,Subtarget);
   case ISD::FMA:            return PerformFMACombine(N, DAG, Subtarget);
   case X86ISD::BLENDI:    return PerformBLENDICombine(N, DAG);
+  case ISD::MGATHER:
+  case ISD::MSCATTER:       return PerformGatherScatterCombine(N, DAG);
   }
 
   return SDValue();

diff --git a/lib/Target/X86/X86InstrAVX512.td b/lib/Target/X86/X86InstrAVX512.td
index 62f28b79ecdb1ec6e07069243eb4b20ef7cec548..48918d66d240da374a3b8646f333bb9ba3b8796f 100644 (file)
--- a/lib/Target/X86/X86InstrAVX512.td
+++ b/lib/Target/X86/X86InstrAVX512.td
@@ -2176,17 +2176,19 @@ let Predicates = [HasAVX512] in {
             (EXTRACT_SUBREG
              (AND32ri (KMOVWrk (COPY_TO_REGCLASS VK1:$src, VK16)), (i32 1)),
               sub_16bit)>;
-  def : Pat<(v16i1 (scalar_to_vector VK1:$src)),
-            (COPY_TO_REGCLASS VK1:$src, VK16)>;
-  def : Pat<(v8i1 (scalar_to_vector VK1:$src)),
-            (COPY_TO_REGCLASS VK1:$src, VK8)>;
-}
-let Predicates = [HasBWI] in {
-  def : Pat<(v32i1 (scalar_to_vector VK1:$src)),
-            (COPY_TO_REGCLASS VK1:$src, VK32)>;
-  def : Pat<(v64i1 (scalar_to_vector VK1:$src)),
-            (COPY_TO_REGCLASS VK1:$src, VK64)>;
 }
+def : Pat<(v16i1 (scalar_to_vector VK1:$src)),
+          (COPY_TO_REGCLASS VK1:$src, VK16)>;
+def : Pat<(v8i1 (scalar_to_vector VK1:$src)),
+          (COPY_TO_REGCLASS VK1:$src, VK8)>;
+def : Pat<(v4i1 (scalar_to_vector VK1:$src)),
+          (COPY_TO_REGCLASS VK1:$src, VK4)>;
+def : Pat<(v2i1 (scalar_to_vector VK1:$src)),
+          (COPY_TO_REGCLASS VK1:$src, VK2)>;
+def : Pat<(v32i1 (scalar_to_vector VK1:$src)),
+          (COPY_TO_REGCLASS VK1:$src, VK32)>;
+def : Pat<(v64i1 (scalar_to_vector VK1:$src)),
+          (COPY_TO_REGCLASS VK1:$src, VK64)>;
 
 
 // With AVX-512 only, 8-bit mask is promoted to 16-bit mask.
@@ -2489,6 +2491,9 @@ def : Pat<(v8i1 (extract_subvector (v16i1 VK16:$src), (iPTR 8))),
 def : Pat<(v16i1 (extract_subvector (v32i1 VK32:$src), (iPTR 0))),
           (v16i1 (COPY_TO_REGCLASS VK32:$src, VK16))>;
 
+def : Pat<(v16i1 (extract_subvector (v32i1 VK32:$src), (iPTR 16))),
+          (v16i1 (COPY_TO_REGCLASS (KSHIFTRDri VK32:$src, (i8 16)), VK16))>;
+
 def : Pat<(v32i1 (extract_subvector (v64i1 VK64:$src), (iPTR 0))),
           (v32i1 (COPY_TO_REGCLASS VK64:$src, VK32))>;
 
@@ -2497,6 +2502,7 @@ def : Pat<(v32i1 (extract_subvector (v64i1 VK64:$src), (iPTR 32))),
 
 def : Pat<(v4i1 (extract_subvector (v8i1 VK8:$src), (iPTR 0))),
           (v4i1 (COPY_TO_REGCLASS VK8:$src, VK4))>;
+
 def : Pat<(v2i1 (extract_subvector (v8i1 VK8:$src), (iPTR 0))),
           (v2i1 (COPY_TO_REGCLASS VK8:$src, VK2))>;
 

diff --git a/test/CodeGen/X86/masked_gather_scatter.ll b/test/CodeGen/X86/masked_gather_scatter.ll
index 19d2c891f85adc27e074857946575d34151af878..84f04c01efc26cb6a6e32b89e7b8d8862655c780 100644 (file)
--- a/test/CodeGen/X86/masked_gather_scatter.ll
+++ b/test/CodeGen/X86/masked_gather_scatter.ll
@@ -1,29 +1,51 @@
-; RUN: llc -mtriple=x86_64-apple-darwin  -mcpu=knl < %s | FileCheck %s -check-prefix=KNL
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu  -mattr=+avx512f < %s | FileCheck %s --check-prefix=KNL_64
+; RUN: llc -mtriple=i386-unknown-linux-gnu  -mattr=+avx512f < %s | FileCheck %s --check-prefix=KNL_32
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu  -mattr=+avx512vl -mattr=+avx512dq < %s | FileCheck %s --check-prefix=SKX
+; RUN: llc -mtriple=i386-unknown-linux-gnu  -mattr=+avx512vl -mattr=+avx512dq < %s | FileCheck %s --check-prefix=SKX_32
 ; RUN: opt -mtriple=x86_64-apple-darwin -codegenprepare -mcpu=corei7-avx -S < %s | FileCheck %s -check-prefix=SCALAR
 
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
-; KNL-LABEL: test1
-; KNL: kxnorw  %k1, %k1, %k1
-; KNL: vgatherdps      (%rdi,%zmm0,4), %zmm1 {%k1}
 
 ; SCALAR-LABEL: test1
-; SCALAR:      extractelement <16 x float*> 
+; SCALAR:      extractelement <16 x float*>
 ; SCALAR-NEXT: load float
 ; SCALAR-NEXT: insertelement <16 x float>
 ; SCALAR-NEXT: extractelement <16 x float*>
 ; SCALAR-NEXT: load float
 
 define <16 x float> @test1(float* %base, <16 x i32> %ind) {
+; KNL_64-LABEL: test1:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_64-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test1:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_32-NEXT:    vgatherdps (%eax,%zmm0,4), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test1:
+; SKX:       # BB#0:
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0
   %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer
 
   %sext_ind = sext <16 x i32> %ind to <16 x i64>
   %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
-  
+
   %res = call <16 x float> @llvm.masked.gather.v16f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
   ret <16 x float>%res
 }
@@ -31,23 +53,41 @@ define <16 x float> @test1(float* %base, <16 x i32> %ind) {
 declare <16 x i32> @llvm.masked.gather.v16i32(<16 x i32*>, i32, <16 x i1>, <16 x i32>)
 declare <16 x float> @llvm.masked.gather.v16f32(<16 x float*>, i32, <16 x i1>, <16 x float>)
 declare <8 x i32> @llvm.masked.gather.v8i32(<8 x i32*> , i32, <8 x i1> , <8 x i32> )
-  
-; KNL-LABEL: test2
-; KNL: kmovw %esi, %k1
-; KNL: vgatherdps      (%rdi,%zmm0,4), %zmm1 {%k1}
+
 
 ; SCALAR-LABEL: test2
-; SCALAR:      extractelement <16 x float*> 
+; SCALAR:      extractelement <16 x float*>
 ; SCALAR-NEXT: load float
 ; SCALAR-NEXT: insertelement <16 x float>
 ; SCALAR-NEXT: br label %else
 ; SCALAR: else:
-; SCALAR-NEXT:  %res.phi.else = phi 
+; SCALAR-NEXT:  %res.phi.else = phi
 ; SCALAR-NEXT:  %Mask1 = extractelement <16 x i1> %imask, i32 1
 ; SCALAR-NEXT:  %ToLoad1 = icmp eq i1 %Mask1, true
 ; SCALAR-NEXT:  br i1 %ToLoad1, label %cond.load1, label %else2
 
 define <16 x float> @test2(float* %base, <16 x i32> %ind, i16 %mask) {
+; KNL_64-LABEL: test2:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kmovw %esi, %k1
+; KNL_64-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test2:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    kmovw {{[0-9]+}}(%esp), %k1
+; KNL_32-NEXT:    vgatherdps (%eax,%zmm0,4), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test2:
+; SKX:       # BB#0:
+; SKX-NEXT:    kmovw %esi, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0
   %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer
@@ -59,10 +99,28 @@ define <16 x float> @test2(float* %base, <16 x i32> %ind, i16 %mask) {
   ret <16 x float> %res
 }
 
-; KNL-LABEL: test3
-; KNL: kmovw %esi, %k1
-; KNL: vpgatherdd      (%rdi,%zmm0,4), %zmm1 {%k1}
 define <16 x i32> @test3(i32* %base, <16 x i32> %ind, i16 %mask) {
+; KNL_64-LABEL: test3:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kmovw %esi, %k1
+; KNL_64-NEXT:    vpgatherdd (%rdi,%zmm0,4), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test3:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    kmovw {{[0-9]+}}(%esp), %k1
+; KNL_32-NEXT:    vpgatherdd (%eax,%zmm0,4), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test3:
+; SKX:       # BB#0:
+; SKX-NEXT:    kmovw %esi, %k1
+; SKX-NEXT:    vpgatherdd (%rdi,%zmm0,4), %zmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <16 x i32*> undef, i32* %base, i32 0
   %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> undef, <16 x i32> zeroinitializer
@@ -74,13 +132,38 @@ define <16 x i32> @test3(i32* %base, <16 x i32> %ind, i16 %mask) {
   ret <16 x i32> %res
 }
 
-; KNL-LABEL: test4
-; KNL: kmovw %esi, %k1
-; KNL: kmovw
-; KNL: vpgatherdd
-; KNL: vpgatherdd
 
 define <16 x i32> @test4(i32* %base, <16 x i32> %ind, i16 %mask) {
+; KNL_64-LABEL: test4:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kmovw %esi, %k1
+; KNL_64-NEXT:    kmovw %k1, %k2
+; KNL_64-NEXT:    vpgatherdd (%rdi,%zmm0,4), %zmm1 {%k2}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm2
+; KNL_64-NEXT:    vpgatherdd (%rdi,%zmm0,4), %zmm2 {%k1}
+; KNL_64-NEXT:    vpaddd %zmm2, %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test4:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    kmovw {{[0-9]+}}(%esp), %k1
+; KNL_32-NEXT:    kmovw %k1, %k2
+; KNL_32-NEXT:    vpgatherdd (%eax,%zmm0,4), %zmm1 {%k2}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm2
+; KNL_32-NEXT:    vpgatherdd (%eax,%zmm0,4), %zmm2 {%k1}
+; KNL_32-NEXT:    vpaddd %zmm2, %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test4:
+; SKX:       # BB#0:
+; SKX-NEXT:    kmovw %esi, %k1
+; SKX-NEXT:    kmovw %k1, %k2
+; SKX-NEXT:    vpgatherdd (%rdi,%zmm0,4), %zmm1 {%k2}
+; SKX-NEXT:    vmovaps %zmm1, %zmm2
+; SKX-NEXT:    vpgatherdd (%rdi,%zmm0,4), %zmm2 {%k1}
+; SKX-NEXT:    vpaddd %zmm2, %zmm1, %zmm0
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <16 x i32*> undef, i32* %base, i32 0
   %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> undef, <16 x i32> zeroinitializer
@@ -93,10 +176,6 @@ define <16 x i32> @test4(i32* %base, <16 x i32> %ind, i16 %mask) {
   ret <16 x i32> %res
 }
 
-; KNL-LABEL: test5
-; KNL: kmovw %k1, %k2
-; KNL: vpscatterdd {{.*}}%k2
-; KNL: vpscatterdd {{.*}}%k1
 
 ; SCALAR-LABEL: test5
 ; SCALAR:        %Mask0 = extractelement <16 x i1> %imask, i32 0
@@ -113,6 +192,30 @@ define <16 x i32> @test4(i32* %base, <16 x i32> %ind, i16 %mask) {
 ; SCALAR-NEXT:  br i1 %ToStore1, label %cond.store1, label %else2
 
 define void @test5(i32* %base, <16 x i32> %ind, i16 %mask, <16 x i32>%val) {
+; KNL_64-LABEL: test5:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kmovw %esi, %k1
+; KNL_64-NEXT:    kmovw %k1, %k2
+; KNL_64-NEXT:    vpscatterdd %zmm1, (%rdi,%zmm0,4) {%k2}
+; KNL_64-NEXT:    vpscatterdd %zmm1, (%rdi,%zmm0,4) {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test5:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    kmovw {{[0-9]+}}(%esp), %k1
+; KNL_32-NEXT:    kmovw %k1, %k2
+; KNL_32-NEXT:    vpscatterdd %zmm1, (%eax,%zmm0,4) {%k2}
+; KNL_32-NEXT:    vpscatterdd %zmm1, (%eax,%zmm0,4) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test5:
+; SKX:       # BB#0:
+; SKX-NEXT:    kmovw %esi, %k1
+; SKX-NEXT:    kmovw %k1, %k2
+; SKX-NEXT:    vpscatterdd %zmm1, (%rdi,%zmm0,4) {%k2}
+; SKX-NEXT:    vpscatterdd %zmm1, (%rdi,%zmm0,4) {%k1}
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <16 x i32*> undef, i32* %base, i32 0
   %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> undef, <16 x i32> zeroinitializer
@@ -127,11 +230,6 @@ define void @test5(i32* %base, <16 x i32> %ind, i16 %mask, <16 x i32>%val) {
 declare void @llvm.masked.scatter.v8i32(<8 x i32> , <8 x i32*> , i32 , <8 x i1> )
 declare void @llvm.masked.scatter.v16i32(<16 x i32> , <16 x i32*> , i32 , <16 x i1> )
 
-; KNL-LABEL: test6
-; KNL: kxnorw  %k1, %k1, %k1
-; KNL: kxnorw  %k2, %k2, %k2
-; KNL: vpgatherqd      (,%zmm{{.*}}), %ymm{{.*}} {%k2}
-; KNL: vpscatterqd     %ymm{{.*}}, (,%zmm{{.*}}) {%k1}
 
 ; SCALAR-LABEL: test6
 ; SCALAR:        store i32 %Elt0, i32* %Ptr01, align 4
@@ -143,6 +241,33 @@ declare void @llvm.masked.scatter.v16i32(<16 x i32> , <16 x i32*> , i32 , <16 x
 ; SCALAR-NEXT:   store i32 %Elt2, i32* %Ptr23, align 4
 
 define <8 x i32> @test6(<8 x i32>%a1, <8 x i32*> %ptr) {
+; KNL_64-LABEL: test6:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_64-NEXT:    kxnorw %k2, %k2, %k2
+; KNL_64-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k2}
+; KNL_64-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; KNL_64-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test6:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_32-NEXT:    vpmovsxdq %ymm1, %zmm2
+; KNL_32-NEXT:    kxnorw %k2, %k2, %k2
+; KNL_32-NEXT:    vpgatherqd (,%zmm2), %ymm1 {%k2}
+; KNL_32-NEXT:    vpscatterqd %ymm0, (,%zmm2) {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test6:
+; SKX:       # BB#0:
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    kxnorw %k2, %k2, %k2
+; SKX-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k2}
+; SKX-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; SKX-NEXT:    vmovaps %zmm2, %zmm0
+; SKX-NEXT:    retq
 
   %a = call <8 x i32> @llvm.masked.gather.v8i32(<8 x i32*> %ptr, i32 4, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i32> undef)
 
@@ -150,13 +275,41 @@ define <8 x i32> @test6(<8 x i32>%a1, <8 x i32*> %ptr) {
   ret <8 x i32>%a
 }
 
-; In this case the index should be promoted to <8 x i64> for KNL
-; KNL-LABEL: test7
-; KNL: vpmovsxdq %ymm0, %zmm0
-; KNL: kmovw   %k1, %k2
-; KNL: vpgatherqd {{.*}} {%k2}
-; KNL: vpgatherqd {{.*}} {%k1}
 define <8 x i32> @test7(i32* %base, <8 x i32> %ind, i8 %mask) {
+;
+; KNL_64-LABEL: test7:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    movzbl %sil, %eax
+; KNL_64-NEXT:    kmovw %eax, %k1
+; KNL_64-NEXT:    vpmovsxdq %ymm0, %zmm0
+; KNL_64-NEXT:    kmovw %k1, %k2
+; KNL_64-NEXT:    vpgatherqd (%rdi,%zmm0,4), %ymm1 {%k2}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm2
+; KNL_64-NEXT:    vpgatherqd (%rdi,%zmm0,4), %ymm2 {%k1}
+; KNL_64-NEXT:    vpaddd %ymm2, %ymm1, %ymm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test7:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    kmovw {{[0-9]+}}(%esp), %k1
+; KNL_32-NEXT:    vpmovsxdq %ymm0, %zmm0
+; KNL_32-NEXT:    kmovw %k1, %k2
+; KNL_32-NEXT:    vpgatherqd (%eax,%zmm0,4), %ymm1 {%k2}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm2
+; KNL_32-NEXT:    vpgatherqd (%eax,%zmm0,4), %ymm2 {%k1}
+; KNL_32-NEXT:    vpaddd %ymm2, %ymm1, %ymm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test7:
+; SKX:       # BB#0:
+; SKX-NEXT:    kmovb %esi, %k1
+; SKX-NEXT:    kmovw %k1, %k2
+; SKX-NEXT:    vpgatherdd (%rdi,%ymm0,4), %ymm1 {%k2}
+; SKX-NEXT:    vmovaps %zmm1, %zmm2
+; SKX-NEXT:    vpgatherdd (%rdi,%ymm0,4), %ymm2 {%k1}
+; SKX-NEXT:    vpaddd %ymm2, %ymm1, %ymm0
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <8 x i32*> undef, i32* %base, i32 0
   %broadcast.splat = shufflevector <8 x i32*> %broadcast.splatinsert, <8 x i32*> undef, <8 x i32> zeroinitializer
@@ -171,15 +324,57 @@ define <8 x i32> @test7(i32* %base, <8 x i32> %ind, i8 %mask) {
 
 ; No uniform base in this case, index <8 x i64> contains addresses,
 ; each gather call will be split into two
-; KNL-LABEL: test8
-; KNL: kshiftrw        $8, %k1, %k2
-; KNL: vpgatherqd
-; KNL: vpgatherqd
-; KNL: vinserti64x4
-; KNL: vpgatherqd
-; KNL: vpgatherqd
-; KNL: vinserti64x4
 define <16 x i32> @test8(<16 x i32*> %ptr.random, <16 x i32> %ind, i16 %mask) {
+; KNL_64-LABEL: test8:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kmovw %edi, %k1
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    kmovw %k2, %k3
+; KNL_64-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k3}
+; KNL_64-NEXT:    kmovw %k1, %k3
+; KNL_64-NEXT:    vpgatherqd (,%zmm0), %ymm3 {%k3}
+; KNL_64-NEXT:    vinserti64x4 $1, %ymm2, %zmm3, %zmm4
+; KNL_64-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k2}
+; KNL_64-NEXT:    vpgatherqd (,%zmm0), %ymm3 {%k1}
+; KNL_64-NEXT:    vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; KNL_64-NEXT:    vpaddd %zmm0, %zmm4, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test8:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    kmovw {{[0-9]+}}(%esp), %k1
+; KNL_32-NEXT:    kmovw %k1, %k2
+; KNL_32-NEXT:    vpgatherdd (,%zmm0), %zmm1 {%k2}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm2
+; KNL_32-NEXT:    vpgatherdd (,%zmm0), %zmm2 {%k1}
+; KNL_32-NEXT:    vpaddd %zmm2, %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test8:
+; SKX:       # BB#0:
+; SKX-NEXT:    kmovw %edi, %k1
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    kmovw %k2, %k3
+; SKX-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k3}
+; SKX-NEXT:    kmovw %k1, %k3
+; SKX-NEXT:    vpgatherqd (,%zmm0), %ymm3 {%k3}
+; SKX-NEXT:    vinserti32x8 $1, %ymm2, %zmm3, %zmm4
+; SKX-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k2}
+; SKX-NEXT:    vpgatherqd (,%zmm0), %ymm3 {%k1}
+; SKX-NEXT:    vinserti32x8 $1, %ymm2, %zmm3, %zmm0
+; SKX-NEXT:    vpaddd %zmm0, %zmm4, %zmm0
+; SKX-NEXT:    retq
+;
+; SKX_32-LABEL: test8:
+; SKX_32:       # BB#0:
+; SKX_32-NEXT:    kmovw {{[0-9]+}}(%esp), %k1
+; SKX_32-NEXT:    kmovw %k1, %k2
+; SKX_32-NEXT:    vpgatherdd (,%zmm0), %zmm1 {%k2}
+; SKX_32-NEXT:    vmovaps %zmm1, %zmm2
+; SKX_32-NEXT:    vpgatherdd (,%zmm0), %zmm2 {%k1}
+; SKX_32-NEXT:    vpaddd %zmm2, %zmm1, %zmm0
+; SKX_32-NEXT:    retl
+
   %imask = bitcast i16 %mask to <16 x i1>
   %gt1 = call <16 x i32> @llvm.masked.gather.v16i32(<16 x i32*> %ptr.random, i32 4, <16 x i1> %imask, <16 x i32>undef)
   %gt2 = call <16 x i32> @llvm.masked.gather.v16i32(<16 x i32*> %ptr.random, i32 4, <16 x i1> %imask, <16 x i32>%gt1)
@@ -193,18 +388,60 @@ define <16 x i32> @test8(<16 x i32*> %ptr.random, <16 x i32> %ind, i16 %mask) {
 ; Masked gather for agregate types
 ; Test9 and Test10 should give the same result (scalar and vector indices in GEP)
 
-; KNL-LABEL: test9
-; KNL: vpbroadcastq    %rdi, %zmm
-; KNL: vpmovsxdq
-; KNL: vpbroadcastq
-; KNL: vpmuludq
-; KNL: vpaddq
-; KNL: vpaddq
-; KNL: vpaddq
-; KNL: vpaddq
-; KNL: vpgatherqd      (,%zmm
 
 define <8 x i32> @test9(%struct.ST* %base, <8 x i64> %ind1, <8 x i32>%ind5) {
+; KNL_64-LABEL: test9:
+; KNL_64:       # BB#0: # %entry
+; KNL_64-NEXT:    vpbroadcastq %rdi, %zmm2
+; KNL_64-NEXT:    vpmovsxdq %ymm1, %zmm1
+; KNL_64-NEXT:    vpbroadcastq {{.*}}(%rip), %zmm3
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm1, %zmm4
+; KNL_64-NEXT:    vpsrlq $32, %zmm1, %zmm1
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm1, %zmm1
+; KNL_64-NEXT:    vpsllq $32, %zmm1, %zmm1
+; KNL_64-NEXT:    vpaddq %zmm1, %zmm4, %zmm1
+; KNL_64-NEXT:    vpbroadcastq {{.*}}(%rip), %zmm3
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm0, %zmm4
+; KNL_64-NEXT:    vpsrlq $32, %zmm0, %zmm0
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm0, %zmm0
+; KNL_64-NEXT:    vpsllq $32, %zmm0, %zmm0
+; KNL_64-NEXT:    vpaddq %zmm0, %zmm4, %zmm0
+; KNL_64-NEXT:    vpaddq %zmm0, %zmm2, %zmm0
+; KNL_64-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; KNL_64-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm1
+; KNL_64-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_64-NEXT:    vpgatherqd (,%zmm1), %ymm0 {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test9:
+; KNL_32:       # BB#0: # %entry
+; KNL_32-NEXT:    vpbroadcastd {{[0-9]+}}(%esp), %ymm2
+; KNL_32-NEXT:    vpbroadcastd .LCPI8_0, %ymm3
+; KNL_32-NEXT:    vpmulld %ymm3, %ymm1, %ymm1
+; KNL_32-NEXT:    vpmovqd %zmm0, %ymm0
+; KNL_32-NEXT:    vpbroadcastd .LCPI8_1, %ymm3
+; KNL_32-NEXT:    vpmulld %ymm3, %ymm0, %ymm0
+; KNL_32-NEXT:    vpaddd %ymm0, %ymm2, %ymm0
+; KNL_32-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
+; KNL_32-NEXT:    vpbroadcastd .LCPI8_2, %ymm1
+; KNL_32-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
+; KNL_32-NEXT:    vpmovsxdq %ymm0, %zmm1
+; KNL_32-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_32-NEXT:    vpgatherqd (,%zmm1), %ymm0 {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test9:
+; SKX:       # BB#0: # %entry
+; SKX-NEXT:    vpbroadcastq %rdi, %zmm2
+; SKX-NEXT:    vpmullq {{.*}}(%rip){1to8}, %zmm0, %zmm0
+; SKX-NEXT:    vpaddq %zmm0, %zmm2, %zmm0
+; SKX-NEXT:    vpmovsxdq %ymm1, %zmm1
+; SKX-NEXT:    vpmullq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; SKX-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm1
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    vpgatherqd (,%zmm1), %ymm0 {%k1}
+; SKX-NEXT:    retq
 entry:
   %broadcast.splatinsert = insertelement <8 x %struct.ST*> undef, %struct.ST* %base, i32 0
   %broadcast.splat = shufflevector <8 x %struct.ST*> %broadcast.splatinsert, <8 x %struct.ST*> undef, <8 x i32> zeroinitializer
@@ -214,17 +451,59 @@ entry:
   ret <8 x i32> %res
 }
 
-; KNL-LABEL: test10
-; KNL: vpbroadcastq    %rdi, %zmm
-; KNL: vpmovsxdq
-; KNL: vpbroadcastq
-; KNL: vpmuludq
-; KNL: vpaddq
-; KNL: vpaddq
-; KNL: vpaddq
-; KNL: vpaddq
-; KNL: vpgatherqd      (,%zmm
 define <8 x i32> @test10(%struct.ST* %base, <8 x i64> %i1, <8 x i32>%ind5) {
+; KNL_64-LABEL: test10:
+; KNL_64:       # BB#0: # %entry
+; KNL_64-NEXT:    vpbroadcastq %rdi, %zmm2
+; KNL_64-NEXT:    vpmovsxdq %ymm1, %zmm1
+; KNL_64-NEXT:    vpbroadcastq {{.*}}(%rip), %zmm3
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm1, %zmm4
+; KNL_64-NEXT:    vpsrlq $32, %zmm1, %zmm1
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm1, %zmm1
+; KNL_64-NEXT:    vpsllq $32, %zmm1, %zmm1
+; KNL_64-NEXT:    vpaddq %zmm1, %zmm4, %zmm1
+; KNL_64-NEXT:    vpbroadcastq {{.*}}(%rip), %zmm3
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm0, %zmm4
+; KNL_64-NEXT:    vpsrlq $32, %zmm0, %zmm0
+; KNL_64-NEXT:    vpmuludq %zmm3, %zmm0, %zmm0
+; KNL_64-NEXT:    vpsllq $32, %zmm0, %zmm0
+; KNL_64-NEXT:    vpaddq %zmm0, %zmm4, %zmm0
+; KNL_64-NEXT:    vpaddq %zmm0, %zmm2, %zmm0
+; KNL_64-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; KNL_64-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm1
+; KNL_64-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_64-NEXT:    vpgatherqd (,%zmm1), %ymm0 {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test10:
+; KNL_32:       # BB#0: # %entry
+; KNL_32-NEXT:    vpbroadcastd {{[0-9]+}}(%esp), %ymm2
+; KNL_32-NEXT:    vpbroadcastd .LCPI9_0, %ymm3
+; KNL_32-NEXT:    vpmulld %ymm3, %ymm1, %ymm1
+; KNL_32-NEXT:    vpmovqd %zmm0, %ymm0
+; KNL_32-NEXT:    vpbroadcastd .LCPI9_1, %ymm3
+; KNL_32-NEXT:    vpmulld %ymm3, %ymm0, %ymm0
+; KNL_32-NEXT:    vpaddd %ymm0, %ymm2, %ymm0
+; KNL_32-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
+; KNL_32-NEXT:    vpbroadcastd .LCPI9_2, %ymm1
+; KNL_32-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
+; KNL_32-NEXT:    vpmovsxdq %ymm0, %zmm1
+; KNL_32-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_32-NEXT:    vpgatherqd (,%zmm1), %ymm0 {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test10:
+; SKX:       # BB#0: # %entry
+; SKX-NEXT:    vpbroadcastq %rdi, %zmm2
+; SKX-NEXT:    vpmullq {{.*}}(%rip){1to8}, %zmm0, %zmm0
+; SKX-NEXT:    vpaddq %zmm0, %zmm2, %zmm0
+; SKX-NEXT:    vpmovsxdq %ymm1, %zmm1
+; SKX-NEXT:    vpmullq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; SKX-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm1
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    vpgatherqd (,%zmm1), %ymm0 {%k1}
+; SKX-NEXT:    retq
 entry:
   %broadcast.splatinsert = insertelement <8 x %struct.ST*> undef, %struct.ST* %base, i32 0
   %broadcast.splat = shufflevector <8 x %struct.ST*> %broadcast.splatinsert, <8 x %struct.ST*> undef, <8 x i32> zeroinitializer
@@ -235,10 +514,28 @@ entry:
 }
 
 ; Splat index in GEP, requires broadcast
-; KNL-LABEL: test11
-; KNL: vpbroadcastd    %esi, %zmm
-; KNL: vgatherdps      (%rdi,%zmm
 define <16 x float> @test11(float* %base, i32 %ind) {
+; KNL_64-LABEL: test11:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpbroadcastd %esi, %zmm1
+; KNL_64-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_64-NEXT:    vgatherdps (%rdi,%zmm1,4), %zmm0 {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test11:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpbroadcastd {{[0-9]+}}(%esp), %zmm1
+; KNL_32-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_32-NEXT:    vgatherdps (%eax,%zmm1,4), %zmm0 {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test11:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpbroadcastd %esi, %zmm1
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%zmm1,4), %zmm0 {%k1}
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0
   %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer
@@ -250,10 +547,28 @@ define <16 x float> @test11(float* %base, i32 %ind) {
 }
 
 ; We are checking the uniform base here. It is taken directly from input to vgatherdps
-; KNL-LABEL: test12
-; KNL: kxnorw  %k1, %k1, %k1
-; KNL: vgatherdps      (%rdi,%zmm
 define <16 x float> @test12(float* %base, <16 x i32> %ind) {
+; KNL_64-LABEL: test12:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_64-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test12:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_32-NEXT:    vgatherdps (%eax,%zmm0,4), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test12:
+; SKX:       # BB#0:
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
 
   %sext_ind = sext <16 x i32> %ind to <16 x i64>
   %gep.random = getelementptr float, float *%base, <16 x i64> %sext_ind
@@ -263,10 +578,25 @@ define <16 x float> @test12(float* %base, <16 x i32> %ind) {
 }
 
 ; The same as the previous, but the mask is undefined
-; KNL-LABEL: test13
-; KNL-NOT: kxnorw
-; KNL: vgatherdps      (%rdi,%zmm
 define <16 x float> @test13(float* %base, <16 x i32> %ind) {
+; KNL_64-LABEL: test13:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test13:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vgatherdps (%eax,%zmm0,4), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test13:
+; SKX:       # BB#0:
+; SKX-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
 
   %sext_ind = sext <16 x i32> %ind to <16 x i64>
   %gep.random = getelementptr float, float *%base, <16 x i64> %sext_ind
@@ -276,10 +606,58 @@ define <16 x float> @test13(float* %base, <16 x i32> %ind) {
 }
 
 ; The base pointer is not splat, can't find unform base
-; KNL-LABEL: test14
-; KNL: vgatherqps      (,%zmm0)
-; KNL: vgatherqps      (,%zmm0)
 define <16 x float> @test14(float* %base, i32 %ind, <16 x float*> %vec) {
+; KNL_64-LABEL: test14:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpinsrq $1, %rdi, %xmm0, %xmm1
+; KNL_64-NEXT:    vinserti32x4 $0, %xmm1, %zmm0, %zmm0
+; KNL_64-NEXT:    vpbroadcastq %xmm0, %zmm0
+; KNL_64-NEXT:    vmovd %esi, %xmm1
+; KNL_64-NEXT:    vpbroadcastd %xmm1, %ymm1
+; KNL_64-NEXT:    vpmovsxdq %ymm1, %zmm1
+; KNL_64-NEXT:    vpsllq $2, %zmm1, %zmm1
+; KNL_64-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; KNL_64-NEXT:    kshiftrw $8, %k0, %k1
+; KNL_64-NEXT:    vgatherqps (,%zmm0), %ymm1 {%k1}
+; KNL_64-NEXT:    vgatherqps (,%zmm0), %ymm2 {%k1}
+; KNL_64-NEXT:    vinsertf64x4 $1, %ymm1, %zmm2, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test14:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpinsrd $1, {{[0-9]+}}(%esp), %xmm0, %xmm1
+; KNL_32-NEXT:    vinserti32x4 $0, %xmm1, %zmm0, %zmm0
+; KNL_32-NEXT:    vpbroadcastd %xmm0, %zmm0
+; KNL_32-NEXT:    vpslld $2, {{[0-9]+}}(%esp){1to16}, %zmm1
+; KNL_32-NEXT:    vpaddd %zmm1, %zmm0, %zmm1
+; KNL_32-NEXT:    vgatherdps (,%zmm1), %zmm0 {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test14:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpinsrq $1, %rdi, %xmm0, %xmm1
+; SKX-NEXT:    vinserti64x2 $0, %xmm1, %zmm0, %zmm0
+; SKX-NEXT:    vpbroadcastq %xmm0, %zmm0
+; SKX-NEXT:    vmovd %esi, %xmm1
+; SKX-NEXT:    vpbroadcastd %xmm1, %ymm1
+; SKX-NEXT:    vpmovsxdq %ymm1, %zmm1
+; SKX-NEXT:    vpsllq $2, %zmm1, %zmm1
+; SKX-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; SKX-NEXT:    kshiftrw $8, %k0, %k1
+; SKX-NEXT:    vgatherqps (,%zmm0), %ymm1 {%k1}
+; SKX-NEXT:    vgatherqps (,%zmm0), %ymm2 {%k1}
+; SKX-NEXT:    vinsertf32x8 $1, %ymm1, %zmm2, %zmm0
+; SKX-NEXT:    retq
+;
+; SKX_32-LABEL: test14:
+; SKX_32:       # BB#0:
+; SKX_32-NEXT:    vpinsrd $1, {{[0-9]+}}(%esp), %xmm0, %xmm1
+; SKX_32-NEXT:    vinserti32x4 $0, %xmm1, %zmm0, %zmm0
+; SKX_32-NEXT:    vpbroadcastd %xmm0, %zmm0
+; SKX_32-NEXT:    vpslld $2, {{[0-9]+}}(%esp){1to16}, %zmm1
+; SKX_32-NEXT:    vpaddd %zmm1, %zmm0, %zmm1
+; SKX_32-NEXT:    vgatherdps (,%zmm1), %zmm0 {%k1}
+; SKX_32-NEXT:    retl
 
   %broadcast.splatinsert = insertelement <16 x float*> %vec, float* %base, i32 1
   %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer
@@ -290,19 +668,585 @@ define <16 x float> @test14(float* %base, i32 %ind, <16 x float*> %vec) {
   ret <16 x float>%res
 }
 
+declare <4 x float> @llvm.masked.gather.v4f32(<4 x float*>, i32, <4 x i1>, <4 x float>)
+declare <4 x double> @llvm.masked.gather.v4f64(<4 x double*>, i32, <4 x i1>, <4 x double>)
+declare <2 x double> @llvm.masked.gather.v2f64(<2 x double*>, i32, <2 x i1>, <2 x double>)
+
+; Gather smaller than existing instruction
+define <4 x float> @test15(float* %base, <4 x i32> %ind, <4 x i1> %mask) {
+;
+; KNL_64-LABEL: test15:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpxor %ymm2, %ymm2, %ymm2
+; KNL_64-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5,6,7]
+; KNL_64-NEXT:    vpmovsxdq %ymm0, %zmm2
+; KNL_64-NEXT:    vpmovsxdq %ymm1, %zmm0
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm0, %zmm0
+; KNL_64-NEXT:    vptestmq %zmm0, %zmm0, %k1
+; KNL_64-NEXT:    vgatherqps (%rdi,%zmm2,4), %ymm0 {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test15:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpxor %ymm2, %ymm2, %ymm2
+; KNL_32-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5,6,7]
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpmovsxdq %ymm0, %zmm2
+; KNL_32-NEXT:    vpmovsxdq %ymm1, %zmm0
+; KNL_32-NEXT:    vpandq .LCPI14_0, %zmm0, %zmm0
+; KNL_32-NEXT:    vptestmq %zmm0, %zmm0, %k1
+; KNL_32-NEXT:    vgatherqps (%eax,%zmm2,4), %ymm0 {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test15:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovd2m %xmm1, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%xmm0,4), %xmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
+
+  %sext_ind = sext <4 x i32> %ind to <4 x i64>
+  %gep.random = getelementptr float, float* %base, <4 x i64> %sext_ind
+  %res = call <4 x float> @llvm.masked.gather.v4f32(<4 x float*> %gep.random, i32 4, <4 x i1> %mask, <4 x float> undef)
+  ret <4 x float>%res
+}
+
+; Gather smaller than existing instruction
+define <4 x double> @test16(double* %base, <4 x i32> %ind, <4 x i1> %mask, <4 x double> %src0) {
+;
+; KNL_64-LABEL: test16:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpslld $31, %xmm1, %xmm1
+; KNL_64-NEXT:    vpsrad $31, %xmm1, %xmm1
+; KNL_64-NEXT:    vpmovsxdq %xmm1, %ymm1
+; KNL_64-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_64-NEXT:    vinserti64x4 $0, %ymm1, %zmm3, %zmm1
+; KNL_64-NEXT:    vpmovsxdq %ymm0, %zmm0
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; KNL_64-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_64-NEXT:    vgatherqpd (%rdi,%zmm0,8), %zmm2 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test16:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpslld $31, %xmm1, %xmm1
+; KNL_32-NEXT:    vpsrad $31, %xmm1, %xmm1
+; KNL_32-NEXT:    vpmovsxdq %xmm1, %ymm1
+; KNL_32-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_32-NEXT:    vinserti64x4 $0, %ymm1, %zmm3, %zmm1
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpmovsxdq %ymm0, %zmm0
+; KNL_32-NEXT:    vpandq .LCPI15_0, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vgatherqpd (%eax,%zmm0,8), %zmm2 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test16:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovd2m %xmm1, %k1
+; SKX-NEXT:    vgatherdpd (%rdi,%xmm0,8), %ymm2 {%k1}
+; SKX-NEXT:    vmovaps %zmm2, %zmm0
+; SKX-NEXT:    retq
+
+  %sext_ind = sext <4 x i32> %ind to <4 x i64>
+  %gep.random = getelementptr double, double* %base, <4 x i64> %sext_ind
+  %res = call <4 x double> @llvm.masked.gather.v4f64(<4 x double*> %gep.random, i32 4, <4 x i1> %mask, <4 x double> %src0)
+  ret <4 x double>%res
+}
+
+define <2 x double> @test17(double* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x double> %src0) {
+;
+; KNL_64-LABEL: test17:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_64-NEXT:    vinserti32x4 $0, %xmm1, %zmm3, %zmm1
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; KNL_64-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_64-NEXT:    vgatherqpd (%rdi,%zmm0,8), %zmm2 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test17:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_32-NEXT:    vinserti32x4 $0, %xmm1, %zmm3, %zmm1
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpandq .LCPI16_0, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vgatherqpd (%eax,%zmm0,8), %zmm2 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test17:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovq2m %xmm1, %k1
+; SKX-NEXT:    vgatherqpd (%rdi,%xmm0,8), %xmm2 {%k1}
+; SKX-NEXT:    vmovaps %zmm2, %zmm0
+; SKX-NEXT:    retq
+
+  %sext_ind = sext <2 x i32> %ind to <2 x i64>
+  %gep.random = getelementptr double, double* %base, <2 x i64> %sext_ind
+  %res = call <2 x double> @llvm.masked.gather.v2f64(<2 x double*> %gep.random, i32 4, <2 x i1> %mask, <2 x double> %src0)
+  ret <2 x double>%res
+}
+
+declare void @llvm.masked.scatter.v4i32(<4 x i32> , <4 x i32*> , i32 , <4 x i1> )
+declare void @llvm.masked.scatter.v4f64(<4 x double> , <4 x double*> , i32 , <4 x i1> )
+declare void @llvm.masked.scatter.v2i64(<2 x i64> , <2 x i64*> , i32 , <2 x i1> )
+declare void @llvm.masked.scatter.v2i32(<2 x i32> , <2 x i32*> , i32 , <2 x i1> )
+declare void @llvm.masked.scatter.v2f32(<2 x float> , <2 x float*> , i32 , <2 x i1> )
+
+define void @test18(<4 x i32>%a1, <4 x i32*> %ptr, <4 x i1>%mask) {
+;
+; KNL_64-LABEL: test18:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpxor %ymm3, %ymm3, %ymm3
+; KNL_64-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7]
+; KNL_64-NEXT:    vpmovsxdq %ymm2, %zmm2
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test18:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpxor %ymm3, %ymm3, %ymm3
+; KNL_32-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7]
+; KNL_32-NEXT:    vpmovsxdq %ymm1, %zmm1
+; KNL_32-NEXT:    vpmovsxdq %ymm2, %zmm2
+; KNL_32-NEXT:    vpandq .LCPI17_0, %zmm2, %zmm2
+; KNL_32-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_32-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test18:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovd2m %xmm2, %k1
+; SKX-NEXT:    vpscatterqd %xmm0, (,%ymm1) {%k1}
+; SKX-NEXT:    retq
+  call void @llvm.masked.scatter.v4i32(<4 x i32> %a1, <4 x i32*> %ptr, i32 4, <4 x i1> %mask)
+  ret void
+}
+
+define void @test19(<4 x double>%a1, double* %ptr, <4 x i1>%mask, <4 x i64> %ind) {
+;
+; KNL_64-LABEL: test19:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpslld $31, %xmm1, %xmm1
+; KNL_64-NEXT:    vpsrad $31, %xmm1, %xmm1
+; KNL_64-NEXT:    vpmovsxdq %xmm1, %ymm1
+; KNL_64-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_64-NEXT:    vinserti64x4 $0, %ymm1, %zmm3, %zmm1
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; KNL_64-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_64-NEXT:    vscatterqpd %zmm0, (%rdi,%zmm2,8) {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test19:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpslld $31, %xmm1, %xmm1
+; KNL_32-NEXT:    vpsrad $31, %xmm1, %xmm1
+; KNL_32-NEXT:    vpmovsxdq %xmm1, %ymm1
+; KNL_32-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_32-NEXT:    vinserti64x4 $0, %ymm1, %zmm3, %zmm1
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpandq .LCPI18_0, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vscatterqpd %zmm0, (%eax,%zmm2,8) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test19:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovd2m %xmm1, %k1
+; SKX-NEXT:    vscatterqpd %ymm0, (%rdi,%ymm2,8) {%k1}
+; SKX-NEXT:    retq
+;
+; SKX_32-LABEL: test19:
+; SKX_32:       # BB#0:
+; SKX_32-NEXT:    vpmovd2m %xmm1, %k1
+; SKX_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; SKX_32-NEXT:    vscatterqpd %ymm0, (%eax,%ymm2,8) {%k1}
+; SKX_32-NEXT:    retl
+  %gep = getelementptr double, double* %ptr, <4 x i64> %ind
+  call void @llvm.masked.scatter.v4f64(<4 x double> %a1, <4 x double*> %gep, i32 8, <4 x i1> %mask)
+  ret void
+}
+
+; Data type requires widening
+define void @test20(<2 x float>%a1, <2 x float*> %ptr, <2 x i1> %mask) {
+;
+; KNL_64-LABEL: test20:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; KNL_64-NEXT:    vmovq %xmm2, %xmm2
+; KNL_64-NEXT:    vpxor %ymm3, %ymm3, %ymm3
+; KNL_64-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7]
+; KNL_64-NEXT:    vpmovsxdq %ymm2, %zmm2
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    vscatterqps %ymm0, (,%zmm1) {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test20:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; KNL_32-NEXT:    vmovq %xmm2, %xmm2
+; KNL_32-NEXT:    vpxor %ymm3, %ymm3, %ymm3
+; KNL_32-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7]
+; KNL_32-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; KNL_32-NEXT:    vpmovsxdq %ymm1, %zmm1
+; KNL_32-NEXT:    vpmovsxdq %ymm2, %zmm2
+; KNL_32-NEXT:    vpandq .LCPI19_0, %zmm2, %zmm2
+; KNL_32-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_32-NEXT:    vscatterqps %ymm0, (,%zmm1) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test20:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovq2m %xmm2, %k0
+; SKX-NEXT:    kshiftlw $2, %k0, %k0
+; SKX-NEXT:    kshiftrw $2, %k0, %k1
+; SKX-NEXT:    vscatterqps %xmm0, (,%ymm1) {%k1}
+; SKX-NEXT:    retq
+  call void @llvm.masked.scatter.v2f32(<2 x float> %a1, <2 x float*> %ptr, i32 4, <2 x i1> %mask)
+  ret void
+}
 
-; KNL-LABEL: test15
-; KNL: kmovw %eax, %k1
-; KNL: vgatherdps      (%rdi,%zmm0,4), %zmm1 {%k1}
+; Data type requires promotion
+define void @test21(<2 x i32>%a1, <2 x i32*> %ptr, <2 x i1>%mask) {
+;
+; KNL_64-LABEL: test21:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_64-NEXT:    vinserti32x4 $0, %xmm2, %zmm3, %zmm2
+; KNL_64-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test21:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_32-NEXT:    vinserti32x4 $0, %xmm2, %zmm3, %zmm2
+; KNL_32-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_32-NEXT:    vpandq .LCPI20_0, %zmm2, %zmm2
+; KNL_32-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_32-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test21:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovq2m %xmm2, %k0
+; SKX-NEXT:    kshiftlw $2, %k0, %k0
+; SKX-NEXT:    kshiftrw $2, %k0, %k1
+; SKX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SKX-NEXT:    vpscatterqd %xmm0, (,%ymm1) {%k1}
+; SKX-NEXT:    retq
+  call void @llvm.masked.scatter.v2i32(<2 x i32> %a1, <2 x i32*> %ptr, i32 4, <2 x i1> %mask)
+  ret void
+}
 
-; SCALAR-LABEL: test15
-; SCALAR:      extractelement <16 x float*> 
+; The result type requires widening
+declare <2 x float> @llvm.masked.gather.v2f32(<2 x float*>, i32, <2 x i1>, <2 x float>)
+
+define <2 x float> @test22(float* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x float> %src0) {
+;
+;
+; KNL_64-LABEL: test22:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; KNL_64-NEXT:    vmovq %xmm1, %xmm1
+; KNL_64-NEXT:    vpxor %ymm3, %ymm3, %ymm3
+; KNL_64-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7]
+; KNL_64-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_64-NEXT:    vpmovsxdq %ymm0, %zmm0
+; KNL_64-NEXT:    vpmovsxdq %ymm1, %zmm1
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; KNL_64-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_64-NEXT:    vgatherqps (%rdi,%zmm0,4), %ymm2 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test22:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; KNL_32-NEXT:    vmovq %xmm1, %xmm1
+; KNL_32-NEXT:    vpxor %ymm3, %ymm3, %ymm3
+; KNL_32-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7]
+; KNL_32-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpmovsxdq %ymm0, %zmm0
+; KNL_32-NEXT:    vpmovsxdq %ymm1, %zmm1
+; KNL_32-NEXT:    vpandq .LCPI21_0, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vgatherqps (%eax,%zmm0,4), %ymm2 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test22:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SKX-NEXT:    vpmovq2m %xmm1, %k0
+; SKX-NEXT:    kshiftlw $2, %k0, %k0
+; SKX-NEXT:    kshiftrw $2, %k0, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%xmm0,4), %xmm2 {%k1}
+; SKX-NEXT:    vmovaps %zmm2, %zmm0
+; SKX-NEXT:    retq
+  %sext_ind = sext <2 x i32> %ind to <2 x i64>
+  %gep.random = getelementptr float, float* %base, <2 x i64> %sext_ind
+  %res = call <2 x float> @llvm.masked.gather.v2f32(<2 x float*> %gep.random, i32 4, <2 x i1> %mask, <2 x float> %src0)
+  ret <2 x float>%res
+}
+
+declare <2 x i32> @llvm.masked.gather.v2i32(<2 x i32*>, i32, <2 x i1>, <2 x i32>)
+declare <2 x i64> @llvm.masked.gather.v2i64(<2 x i64*>, i32, <2 x i1>, <2 x i64>)
+
+define <2 x i32> @test23(i32* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x i32> %src0) {
+;
+; KNL_64-LABEL: test23:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_64-NEXT:    vinserti32x4 $0, %xmm1, %zmm3, %zmm1
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; KNL_64-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_64-NEXT:    vpgatherqq (%rdi,%zmm0,8), %zmm2 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test23:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_32-NEXT:    vinserti32x4 $0, %xmm1, %zmm3, %zmm1
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpandq .LCPI22_0, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vpgatherqq (%eax,%zmm0,8), %zmm2 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test23:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovq2m %xmm1, %k1
+; SKX-NEXT:    vpgatherqq (%rdi,%xmm0,8), %xmm2 {%k1}
+; SKX-NEXT:    vmovaps %zmm2, %zmm0
+; SKX-NEXT:    retq
+  %sext_ind = sext <2 x i32> %ind to <2 x i64>
+  %gep.random = getelementptr i32, i32* %base, <2 x i64> %sext_ind
+  %res = call <2 x i32> @llvm.masked.gather.v2i32(<2 x i32*> %gep.random, i32 4, <2 x i1> %mask, <2 x i32> %src0)
+  ret <2 x i32>%res
+}
+
+define <2 x i32> @test24(i32* %base, <2 x i32> %ind) {
+;
+;
+; KNL_64-LABEL: test24:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    movb $3, %al
+; KNL_64-NEXT:    movzbl %al, %eax
+; KNL_64-NEXT:    kmovw %eax, %k1
+; KNL_64-NEXT:    vpgatherqq (%rdi,%zmm0,8), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test24:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpxord %zmm1, %zmm1, %zmm1
+; KNL_32-NEXT:    vinserti32x4 $0, .LCPI23_0, %zmm1, %zmm1
+; KNL_32-NEXT:    vpandq .LCPI23_1, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vpgatherqq (%eax,%zmm0,8), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test24:
+; SKX:       # BB#0:
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    vpgatherqq (%rdi,%xmm0,8), %xmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
+  %sext_ind = sext <2 x i32> %ind to <2 x i64>
+  %gep.random = getelementptr i32, i32* %base, <2 x i64> %sext_ind
+  %res = call <2 x i32> @llvm.masked.gather.v2i32(<2 x i32*> %gep.random, i32 4, <2 x i1> <i1 true, i1 true>, <2 x i32> undef)
+  ret <2 x i32>%res
+}
+
+define <2 x i64> @test25(i64* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x i64> %src0) {
+;
+; KNL_64-LABEL: test25:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_64-NEXT:    vinserti32x4 $0, %xmm1, %zmm3, %zmm1
+; KNL_64-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm1, %zmm1
+; KNL_64-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_64-NEXT:    vpgatherqq (%rdi,%zmm0,8), %zmm2 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test25:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpxord %zmm3, %zmm3, %zmm3
+; KNL_32-NEXT:    vinserti32x4 $0, %xmm1, %zmm3, %zmm1
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpandq .LCPI24_0, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmq %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vpgatherqq (%eax,%zmm0,8), %zmm2 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test25:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovq2m %xmm1, %k1
+; SKX-NEXT:    vpgatherqq (%rdi,%xmm0,8), %xmm2 {%k1}
+; SKX-NEXT:    vmovaps %zmm2, %zmm0
+; SKX-NEXT:    retq
+  %sext_ind = sext <2 x i32> %ind to <2 x i64>
+  %gep.random = getelementptr i64, i64* %base, <2 x i64> %sext_ind
+  %res = call <2 x i64> @llvm.masked.gather.v2i64(<2 x i64*> %gep.random, i32 8, <2 x i1> %mask, <2 x i64> %src0)
+  ret <2 x i64>%res
+}
+
+define <2 x i64> @test26(i64* %base, <2 x i32> %ind, <2 x i64> %src0) {
+;
+; KNL_64-LABEL: test26:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    movb $3, %al
+; KNL_64-NEXT:    movzbl %al, %eax
+; KNL_64-NEXT:    kmovw %eax, %k1
+; KNL_64-NEXT:    vpgatherqq (%rdi,%zmm0,8), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test26:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpxord %zmm2, %zmm2, %zmm2
+; KNL_32-NEXT:    vinserti32x4 $0, .LCPI25_0, %zmm2, %zmm2
+; KNL_32-NEXT:    vpandq .LCPI25_1, %zmm2, %zmm2
+; KNL_32-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_32-NEXT:    vpgatherqq (%eax,%zmm0,8), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test26:
+; SKX:       # BB#0:
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    vpgatherqq (%rdi,%xmm0,8), %xmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
+  %sext_ind = sext <2 x i32> %ind to <2 x i64>
+  %gep.random = getelementptr i64, i64* %base, <2 x i64> %sext_ind
+  %res = call <2 x i64> @llvm.masked.gather.v2i64(<2 x i64*> %gep.random, i32 8, <2 x i1> <i1 true, i1 true>, <2 x i64> %src0)
+  ret <2 x i64>%res
+}
+
+; Result type requires widening; all-ones mask
+define <2 x float> @test27(float* %base, <2 x i32> %ind) {
+;
+; KNL_64-LABEL: test27:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_64-NEXT:    vpmovsxdq %ymm0, %zmm1
+; KNL_64-NEXT:    movb $3, %al
+; KNL_64-NEXT:    movzbl %al, %eax
+; KNL_64-NEXT:    kmovw %eax, %k1
+; KNL_64-NEXT:    vgatherqps (%rdi,%zmm1,4), %ymm0 {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test27:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    vpmovsxdq %ymm0, %zmm1
+; KNL_32-NEXT:    movb $3, %cl
+; KNL_32-NEXT:    movzbl %cl, %ecx
+; KNL_32-NEXT:    kmovw %ecx, %k1
+; KNL_32-NEXT:    vgatherqps (%eax,%zmm1,4), %ymm0 {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test27:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[0,2,2,3]
+; SKX-NEXT:    movb $3, %al
+; SKX-NEXT:    kmovb %eax, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%xmm1,4), %xmm0 {%k1}
+; SKX-NEXT:    retq
+  %sext_ind = sext <2 x i32> %ind to <2 x i64>
+  %gep.random = getelementptr float, float* %base, <2 x i64> %sext_ind
+  %res = call <2 x float> @llvm.masked.gather.v2f32(<2 x float*> %gep.random, i32 4, <2 x i1> <i1 true, i1 true>, <2 x float> undef)
+  ret <2 x float>%res
+}
+
+; Data type requires promotion, mask is all-ones
+define void @test28(<2 x i32>%a1, <2 x i32*> %ptr) {
+;
+;
+; KNL_64-LABEL: test28:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_64-NEXT:    movb $3, %al
+; KNL_64-NEXT:    movzbl %al, %eax
+; KNL_64-NEXT:    kmovw %eax, %k1
+; KNL_64-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test28:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; KNL_32-NEXT:    vpxord %zmm2, %zmm2, %zmm2
+; KNL_32-NEXT:    vinserti32x4 $0, .LCPI27_0, %zmm2, %zmm2
+; KNL_32-NEXT:    vpandq .LCPI27_1, %zmm2, %zmm2
+; KNL_32-NEXT:    vptestmq %zmm2, %zmm2, %k1
+; KNL_32-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test28:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SKX-NEXT:    movb $3, %al
+; SKX-NEXT:    kmovb %eax, %k1
+; SKX-NEXT:    vpscatterqd %xmm0, (,%ymm1) {%k1}
+; SKX-NEXT:    retq
+  call void @llvm.masked.scatter.v2i32(<2 x i32> %a1, <2 x i32*> %ptr, i32 4, <2 x i1> <i1 true, i1 true>)
+  ret void
+}
+
+
+; SCALAR-LABEL: test29
+; SCALAR:      extractelement <16 x float*>
 ; SCALAR-NEXT: load float
 ; SCALAR-NEXT: insertelement <16 x float>
 ; SCALAR-NEXT: extractelement <16 x float*>
 ; SCALAR-NEXT: load float
 
-define <16 x float> @test15(float* %base, <16 x i32> %ind) {
+define <16 x float> @test29(float* %base, <16 x i32> %ind) {
+; KNL_64-LABEL: test29:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    movw $44, %ax
+; KNL_64-NEXT:    kmovw %eax, %k1
+; KNL_64-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test29:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    movw $44, %cx
+; KNL_32-NEXT:    kmovw %ecx, %k1
+; KNL_32-NEXT:    vgatherdps (%eax,%zmm0,4), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test29:
+; SKX:       # BB#0:
+; SKX-NEXT:    movw $44, %ax
+; SKX-NEXT:    kmovw %eax, %k1
+; SKX-NEXT:    vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    retq
 
   %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0
   %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer
@@ -316,14 +1260,136 @@ define <16 x float> @test15(float* %base, <16 x i32> %ind) {
 
 ; Check non-power-of-2 case. It should be scalarized.
 declare <3 x i32> @llvm.masked.gather.v3i32(<3 x i32*>, i32, <3 x i1>, <3 x i32>)
-; KNL-LABEL: test16
-; KNL: testb
-; KNL: je
-; KNL: testb
-; KNL: je
-; KNL: testb
-; KNL: je
-define <3 x i32> @test16(<3 x i32*> %base, <3 x i32> %ind, <3 x i1> %mask, <3 x i32> %src0) {
+define <3 x i32> @test30(<3 x i32*> %base, <3 x i32> %ind, <3 x i1> %mask, <3 x i32> %src0) {
+; KNL_64-LABEL: test30:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    andl $1, %edx
+; KNL_64-NEXT:    kmovw %edx, %k1
+; KNL_64-NEXT:    andl $1, %esi
+; KNL_64-NEXT:    kmovw %esi, %k2
+; KNL_64-NEXT:    movl %edi, %eax
+; KNL_64-NEXT:    andl $1, %eax
+; KNL_64-NEXT:    kmovw %eax, %k0
+; KNL_64-NEXT:    vpmovsxdq %xmm1, %ymm1
+; KNL_64-NEXT:    vpsllq $2, %ymm1, %ymm1
+; KNL_64-NEXT:    vpaddq %ymm1, %ymm0, %ymm1
+; KNL_64-NEXT:    # implicit-def: %XMM0
+; KNL_64-NEXT:    testb $1, %dil
+; KNL_64-NEXT:    je .LBB29_2
+; KNL_64-NEXT:  # BB#1: # %cond.load
+; KNL_64-NEXT:    vmovq %xmm1, %rax
+; KNL_64-NEXT:    vmovd (%rax), %xmm0
+; KNL_64-NEXT:  .LBB29_2: # %else
+; KNL_64-NEXT:    kmovw %k2, %eax
+; KNL_64-NEXT:    movl %eax, %ecx
+; KNL_64-NEXT:    andl $1, %ecx
+; KNL_64-NEXT:    testb %cl, %cl
+; KNL_64-NEXT:    je .LBB29_4
+; KNL_64-NEXT:  # BB#3: # %cond.load1
+; KNL_64-NEXT:    vpextrq $1, %xmm1, %rcx
+; KNL_64-NEXT:    vpinsrd $1, (%rcx), %xmm0, %xmm0
+; KNL_64-NEXT:  .LBB29_4: # %else2
+; KNL_64-NEXT:    kmovw %k1, %ecx
+; KNL_64-NEXT:    movl %ecx, %edx
+; KNL_64-NEXT:    andl $1, %edx
+; KNL_64-NEXT:    testb %dl, %dl
+; KNL_64-NEXT:    je .LBB29_6
+; KNL_64-NEXT:  # BB#5: # %cond.load4
+; KNL_64-NEXT:    vextracti128 $1, %ymm1, %xmm1
+; KNL_64-NEXT:    vmovq %xmm1, %rdx
+; KNL_64-NEXT:    vpinsrd $2, (%rdx), %xmm0, %xmm0
+; KNL_64-NEXT:  .LBB29_6: # %else5
+; KNL_64-NEXT:    kmovw %k0, %edx
+; KNL_64-NEXT:    vmovd %edx, %xmm1
+; KNL_64-NEXT:    vpinsrd $1, %eax, %xmm1, %xmm1
+; KNL_64-NEXT:    vpinsrd $2, %ecx, %xmm1, %xmm1
+; KNL_64-NEXT:    vpslld $31, %xmm1, %xmm1
+; KNL_64-NEXT:    vblendvps %xmm1, %xmm0, %xmm2, %xmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test30:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    andl $1, %eax
+; KNL_32-NEXT:    kmovw %eax, %k1
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    andl $1, %eax
+; KNL_32-NEXT:    kmovw %eax, %k2
+; KNL_32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL_32-NEXT:    movl %eax, %ecx
+; KNL_32-NEXT:    andl $1, %ecx
+; KNL_32-NEXT:    kmovw %ecx, %k0
+; KNL_32-NEXT:    vpslld $2, %xmm1, %xmm1
+; KNL_32-NEXT:    vpaddd %xmm1, %xmm0, %xmm1
+; KNL_32-NEXT:    # implicit-def: %XMM0
+; KNL_32-NEXT:    testb $1, %al
+; KNL_32-NEXT:    je .LBB29_2
+; KNL_32-NEXT:  # BB#1: # %cond.load
+; KNL_32-NEXT:    vmovd %xmm1, %eax
+; KNL_32-NEXT:    vmovd (%eax), %xmm0
+; KNL_32-NEXT:  .LBB29_2: # %else
+; KNL_32-NEXT:    kmovw %k2, %eax
+; KNL_32-NEXT:    movl %eax, %ecx
+; KNL_32-NEXT:    andl $1, %ecx
+; KNL_32-NEXT:    testb %cl, %cl
+; KNL_32-NEXT:    je .LBB29_4
+; KNL_32-NEXT:  # BB#3: # %cond.load1
+; KNL_32-NEXT:    vpextrd $1, %xmm1, %ecx
+; KNL_32-NEXT:    vpinsrd $1, (%ecx), %xmm0, %xmm0
+; KNL_32-NEXT:  .LBB29_4: # %else2
+; KNL_32-NEXT:    kmovw %k1, %ecx
+; KNL_32-NEXT:    movl %ecx, %edx
+; KNL_32-NEXT:    andl $1, %edx
+; KNL_32-NEXT:    testb %dl, %dl
+; KNL_32-NEXT:    je .LBB29_6
+; KNL_32-NEXT:  # BB#5: # %cond.load4
+; KNL_32-NEXT:    vpextrd $2, %xmm1, %edx
+; KNL_32-NEXT:    vpinsrd $2, (%edx), %xmm0, %xmm0
+; KNL_32-NEXT:  .LBB29_6: # %else5
+; KNL_32-NEXT:    kmovw %k0, %edx
+; KNL_32-NEXT:    vmovd %edx, %xmm1
+; KNL_32-NEXT:    vpinsrd $1, %eax, %xmm1, %xmm1
+; KNL_32-NEXT:    vpinsrd $2, %ecx, %xmm1, %xmm1
+; KNL_32-NEXT:    vpslld $31, %xmm1, %xmm1
+; KNL_32-NEXT:    vblendvps %xmm1, %xmm0, %xmm2, %xmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test30:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovd2m %xmm2, %k1
+; SKX-NEXT:    kmovb %k1, -{{[0-9]+}}(%rsp)
+; SKX-NEXT:    vpmovsxdq %xmm1, %ymm1
+; SKX-NEXT:    vpsllq $2, %ymm1, %ymm1
+; SKX-NEXT:    vpaddq %ymm1, %ymm0, %ymm1
+; SKX-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SKX-NEXT:    # implicit-def: %XMM0
+; SKX-NEXT:    andb $1, %al
+; SKX-NEXT:    je .LBB29_2
+; SKX-NEXT:  # BB#1: # %cond.load
+; SKX-NEXT:    vmovq %xmm1, %rax
+; SKX-NEXT:    vmovd (%rax), %xmm0
+; SKX-NEXT:  .LBB29_2: # %else
+; SKX-NEXT:    kmovb %k1, -{{[0-9]+}}(%rsp)
+; SKX-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SKX-NEXT:    andb $1, %al
+; SKX-NEXT:    je .LBB29_4
+; SKX-NEXT:  # BB#3: # %cond.load1
+; SKX-NEXT:    vpextrq $1, %xmm1, %rax
+; SKX-NEXT:    vpinsrd $1, (%rax), %xmm0, %xmm0
+; SKX-NEXT:  .LBB29_4: # %else2
+; SKX-NEXT:    kmovb %k1, -{{[0-9]+}}(%rsp)
+; SKX-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SKX-NEXT:    andb $1, %al
+; SKX-NEXT:    je .LBB29_6
+; SKX-NEXT:  # BB#5: # %cond.load4
+; SKX-NEXT:    vextracti128 $1, %ymm1, %xmm1
+; SKX-NEXT:    vmovq %xmm1, %rax
+; SKX-NEXT:    vpinsrd $2, (%rax), %xmm0, %xmm0
+; SKX-NEXT:  .LBB29_6: # %else5
+; SKX-NEXT:    vmovdqa32 %xmm0, %xmm3 {%k1}
+; SKX-NEXT:    vmovaps %zmm3, %zmm0
+; SKX-NEXT:    retq
+
   %sext_ind = sext <3 x i32> %ind to <3 x i64>
   %gep.random = getelementptr i32, <3 x i32*> %base, <3 x i64> %sext_ind
   %res = call <3 x i32> @llvm.masked.gather.v3i32(<3 x i32*> %gep.random, i32 4, <3 x i1> %mask, <3 x i32> %src0)
@@ -332,11 +1398,405 @@ define <3 x i32> @test16(<3 x i32*> %base, <3 x i32> %ind, <3 x i1> %mask, <3 x
 
 declare <16 x float*> @llvm.masked.gather.v16p0f32(<16 x float**>, i32, <16 x i1>, <16 x float*>)
 
-; KNL-LABEL: test17
+; KNL-LABEL: test31
 ; KNL: vpgatherqq
 ; KNL: vpgatherqq
-define <16 x float*> @test17(<16 x float**> %ptrs) {
+define <16 x float*> @test31(<16 x float**> %ptrs) {
+; KNL_64-LABEL: test31:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_64-NEXT:    kxnorw %k2, %k2, %k2
+; KNL_64-NEXT:    vpgatherqq (,%zmm0), %zmm2 {%k2}
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k1
+; KNL_64-NEXT:    vpgatherqq (,%zmm1), %zmm3 {%k1}
+; KNL_64-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_64-NEXT:    vmovaps %zmm3, %zmm1
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test31:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    kxnorw %k1, %k1, %k1
+; KNL_32-NEXT:    vpgatherdd (,%zmm0), %zmm1 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm1, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test31:
+; SKX:       # BB#0:
+; SKX-NEXT:    kxnorw %k1, %k1, %k1
+; SKX-NEXT:    kxnorw %k2, %k2, %k2
+; SKX-NEXT:    vpgatherqq (,%zmm0), %zmm2 {%k2}
+; SKX-NEXT:    kshiftrw $8, %k1, %k1
+; SKX-NEXT:    vpgatherqq (,%zmm1), %zmm3 {%k1}
+; SKX-NEXT:    vmovaps %zmm2, %zmm0
+; SKX-NEXT:    vmovaps %zmm3, %zmm1
+; SKX-NEXT:    retq
+;
+; SKX_32-LABEL: test31:
+; SKX_32:       # BB#0:
+; SKX_32-NEXT:    kxnorw %k1, %k1, %k1
+; SKX_32-NEXT:    vpgatherdd (,%zmm0), %zmm1 {%k1}
+; SKX_32-NEXT:    vmovaps %zmm1, %zmm0
+; SKX_32-NEXT:    retl
 
   %res = call <16 x float*> @llvm.masked.gather.v16p0f32(<16 x float**> %ptrs, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float*> undef)
   ret <16 x float*>%res
 }
+
+define <16 x i32> @test_gather_16i32(<16 x i32*> %ptrs, <16 x i1> %mask, <16 x i32> %src0)  {
+; KNL_64-LABEL: test_gather_16i32:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    vextracti64x4 $1, %zmm3, %ymm2
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k2}
+; KNL_64-NEXT:    vpgatherqd (,%zmm0), %ymm3 {%k1}
+; KNL_64-NEXT:    vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_gather_16i32:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI31_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vpgatherdd (,%zmm0), %zmm2 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_gather_16i32:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    vextracti32x8 $1, %zmm3, %ymm2
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vpgatherqd (,%zmm1), %ymm2 {%k2}
+; SKX-NEXT:    vpgatherqd (,%zmm0), %ymm3 {%k1}
+; SKX-NEXT:    vinserti32x8 $1, %ymm2, %zmm3, %zmm0
+; SKX-NEXT:    retq
+;
+; SKX_32-LABEL: test_gather_16i32:
+; SKX_32:       # BB#0:
+; SKX_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; SKX_32-NEXT:    vpandd .LCPI31_0{1to16}, %zmm1, %zmm1
+; SKX_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; SKX_32-NEXT:    vpgatherdd (,%zmm0), %zmm2 {%k1}
+; SKX_32-NEXT:    vmovaps %zmm2, %zmm0
+; SKX_32-NEXT:    retl
+  %res = call <16 x i32> @llvm.masked.gather.v16i32(<16 x i32*> %ptrs, i32 4, <16 x i1> %mask, <16 x i32> %src0)
+  ret <16 x i32> %res
+}
+define <16 x i64> @test_gather_16i64(<16 x i64*> %ptrs, <16 x i1> %mask, <16 x i64> %src0)  {
+; KNL_64-LABEL: test_gather_16i64:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vpgatherqq (,%zmm0), %zmm3 {%k1}
+; KNL_64-NEXT:    vpgatherqq (,%zmm1), %zmm4 {%k2}
+; KNL_64-NEXT:    vmovaps %zmm3, %zmm0
+; KNL_64-NEXT:    vmovaps %zmm4, %zmm1
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_gather_16i64:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    pushl %ebp
+; KNL_32-NEXT:  .Ltmp0:
+; KNL_32-NEXT:    .cfi_def_cfa_offset 8
+; KNL_32-NEXT:  .Ltmp1:
+; KNL_32-NEXT:    .cfi_offset %ebp, -8
+; KNL_32-NEXT:    movl %esp, %ebp
+; KNL_32-NEXT:  .Ltmp2:
+; KNL_32-NEXT:    .cfi_def_cfa_register %ebp
+; KNL_32-NEXT:    andl $-64, %esp
+; KNL_32-NEXT:    subl $64, %esp
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI32_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vmovdqa64 8(%ebp), %zmm1
+; KNL_32-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_32-NEXT:    vpgatherdq (,%ymm0), %zmm2 {%k1}
+; KNL_32-NEXT:    vextracti64x4 $1, %zmm0, %ymm0
+; KNL_32-NEXT:    vpgatherdq (,%ymm0), %zmm1 {%k2}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    movl %ebp, %esp
+; KNL_32-NEXT:    popl %ebp
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_gather_16i64:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vpgatherqq (,%zmm0), %zmm3 {%k1}
+; SKX-NEXT:    vpgatherqq (,%zmm1), %zmm4 {%k2}
+; SKX-NEXT:    vmovaps %zmm3, %zmm0
+; SKX-NEXT:    vmovaps %zmm4, %zmm1
+; SKX-NEXT:    retq
+  %res = call <16 x i64> @llvm.masked.gather.v16i64(<16 x i64*> %ptrs, i32 4, <16 x i1> %mask, <16 x i64> %src0)
+  ret <16 x i64> %res
+}
+declare <16 x i64> @llvm.masked.gather.v16i64(<16 x i64*> %ptrs, i32, <16 x i1> %mask, <16 x i64> %src0)
+define <16 x float> @test_gather_16f32(<16 x float*> %ptrs, <16 x i1> %mask, <16 x float> %src0)  {
+; KNL_64-LABEL: test_gather_16f32:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    vextractf64x4 $1, %zmm3, %ymm2
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vgatherqps (,%zmm1), %ymm2 {%k2}
+; KNL_64-NEXT:    vgatherqps (,%zmm0), %ymm3 {%k1}
+; KNL_64-NEXT:    vinsertf64x4 $1, %ymm2, %zmm3, %zmm0
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_gather_16f32:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI33_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vgatherdps (,%zmm0), %zmm2 {%k1}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_gather_16f32:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    vextractf32x8 $1, %zmm3, %ymm2
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vgatherqps (,%zmm1), %ymm2 {%k2}
+; SKX-NEXT:    vgatherqps (,%zmm0), %ymm3 {%k1}
+; SKX-NEXT:    vinsertf32x8 $1, %ymm2, %zmm3, %zmm0
+; SKX-NEXT:    retq
+  %res = call <16 x float> @llvm.masked.gather.v16f32(<16 x float*> %ptrs, i32 4, <16 x i1> %mask, <16 x float> %src0)
+  ret <16 x float> %res
+}
+define <16 x double> @test_gather_16f64(<16 x double*> %ptrs, <16 x i1> %mask, <16 x double> %src0)  {
+; KNL_64-LABEL: test_gather_16f64:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vgatherqpd (,%zmm0), %zmm3 {%k1}
+; KNL_64-NEXT:    vgatherqpd (,%zmm1), %zmm4 {%k2}
+; KNL_64-NEXT:    vmovaps %zmm3, %zmm0
+; KNL_64-NEXT:    vmovaps %zmm4, %zmm1
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_gather_16f64:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    pushl %ebp
+; KNL_32-NEXT:  .Ltmp3:
+; KNL_32-NEXT:    .cfi_def_cfa_offset 8
+; KNL_32-NEXT:  .Ltmp4:
+; KNL_32-NEXT:    .cfi_offset %ebp, -8
+; KNL_32-NEXT:    movl %esp, %ebp
+; KNL_32-NEXT:  .Ltmp5:
+; KNL_32-NEXT:    .cfi_def_cfa_register %ebp
+; KNL_32-NEXT:    andl $-64, %esp
+; KNL_32-NEXT:    subl $64, %esp
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI34_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vmovapd 8(%ebp), %zmm1
+; KNL_32-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_32-NEXT:    vgatherdpd (,%ymm0), %zmm2 {%k1}
+; KNL_32-NEXT:    vextracti64x4 $1, %zmm0, %ymm0
+; KNL_32-NEXT:    vgatherdpd (,%ymm0), %zmm1 {%k2}
+; KNL_32-NEXT:    vmovaps %zmm2, %zmm0
+; KNL_32-NEXT:    movl %ebp, %esp
+; KNL_32-NEXT:    popl %ebp
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_gather_16f64:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vgatherqpd (,%zmm0), %zmm3 {%k1}
+; SKX-NEXT:    vgatherqpd (,%zmm1), %zmm4 {%k2}
+; SKX-NEXT:    vmovaps %zmm3, %zmm0
+; SKX-NEXT:    vmovaps %zmm4, %zmm1
+; SKX-NEXT:    retq
+  %res = call <16 x double> @llvm.masked.gather.v16f64(<16 x double*> %ptrs, i32 4, <16 x i1> %mask, <16 x double> %src0)
+  ret <16 x double> %res
+}
+declare <16 x double> @llvm.masked.gather.v16f64(<16 x double*> %ptrs, i32, <16 x i1> %mask, <16 x double> %src0)
+define void @test_scatter_16i32(<16 x i32*> %ptrs, <16 x i1> %mask, <16 x i32> %src0)  {
+; KNL_64-LABEL: test_scatter_16i32:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vpscatterqd %ymm3, (,%zmm0) {%k1}
+; KNL_64-NEXT:    vextracti64x4 $1, %zmm3, %ymm0
+; KNL_64-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k2}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_scatter_16i32:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI35_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vpscatterdd %zmm2, (,%zmm0) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_scatter_16i32:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vpscatterqd %ymm3, (,%zmm0) {%k1}
+; SKX-NEXT:    vextracti32x8 $1, %zmm3, %ymm0
+; SKX-NEXT:    vpscatterqd %ymm0, (,%zmm1) {%k2}
+; SKX-NEXT:    retq
+;
+; SKX_32-LABEL: test_scatter_16i32:
+; SKX_32:       # BB#0:
+; SKX_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; SKX_32-NEXT:    vpandd .LCPI35_0{1to16}, %zmm1, %zmm1
+; SKX_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; SKX_32-NEXT:    vpscatterdd %zmm2, (,%zmm0) {%k1}
+; SKX_32-NEXT:    retl
+  call void @llvm.masked.scatter.v16i32(<16 x i32> %src0, <16 x i32*> %ptrs, i32 4, <16 x i1> %mask)
+  ret void
+}
+define void @test_scatter_16i64(<16 x i64*> %ptrs, <16 x i1> %mask, <16 x i64> %src0)  {
+; KNL_64-LABEL: test_scatter_16i64:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vpscatterqq %zmm3, (,%zmm0) {%k1}
+; KNL_64-NEXT:    vpscatterqq %zmm4, (,%zmm1) {%k2}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_scatter_16i64:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    pushl %ebp
+; KNL_32-NEXT:  .Ltmp6:
+; KNL_32-NEXT:    .cfi_def_cfa_offset 8
+; KNL_32-NEXT:  .Ltmp7:
+; KNL_32-NEXT:    .cfi_offset %ebp, -8
+; KNL_32-NEXT:    movl %esp, %ebp
+; KNL_32-NEXT:  .Ltmp8:
+; KNL_32-NEXT:    .cfi_def_cfa_register %ebp
+; KNL_32-NEXT:    andl $-64, %esp
+; KNL_32-NEXT:    subl $64, %esp
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI36_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vmovdqa64 8(%ebp), %zmm1
+; KNL_32-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_32-NEXT:    vpscatterdq %zmm2, (,%ymm0) {%k1}
+; KNL_32-NEXT:    vextracti64x4 $1, %zmm0, %ymm0
+; KNL_32-NEXT:    vpscatterdq %zmm1, (,%ymm0) {%k2}
+; KNL_32-NEXT:    movl %ebp, %esp
+; KNL_32-NEXT:    popl %ebp
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_scatter_16i64:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vpscatterqq %zmm3, (,%zmm0) {%k1}
+; SKX-NEXT:    vpscatterqq %zmm4, (,%zmm1) {%k2}
+; SKX-NEXT:    retq
+  call void @llvm.masked.scatter.v16i64(<16 x i64> %src0, <16 x i64*> %ptrs, i32 4, <16 x i1> %mask)
+  ret void
+}
+declare void @llvm.masked.scatter.v16i64(<16 x i64> %src0, <16 x i64*> %ptrs, i32, <16 x i1> %mask)
+define void @test_scatter_16f32(<16 x float*> %ptrs, <16 x i1> %mask, <16 x float> %src0)  {
+; KNL_64-LABEL: test_scatter_16f32:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vscatterqps %ymm3, (,%zmm0) {%k1}
+; KNL_64-NEXT:    vextractf64x4 $1, %zmm3, %ymm0
+; KNL_64-NEXT:    vscatterqps %ymm0, (,%zmm1) {%k2}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_scatter_16f32:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI37_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vscatterdps %zmm2, (,%zmm0) {%k1}
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_scatter_16f32:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vscatterqps %ymm3, (,%zmm0) {%k1}
+; SKX-NEXT:    vextractf32x8 $1, %zmm3, %ymm0
+; SKX-NEXT:    vscatterqps %ymm0, (,%zmm1) {%k2}
+; SKX-NEXT:    retq
+  call void @llvm.masked.scatter.v16f32(<16 x float> %src0, <16 x float*> %ptrs, i32 4, <16 x i1> %mask)
+  ret void
+}
+declare void @llvm.masked.scatter.v16f32(<16 x float> %src0, <16 x float*> %ptrs, i32, <16 x i1> %mask)
+define void @test_scatter_16f64(<16 x double*> %ptrs, <16 x i1> %mask, <16 x double> %src0)  {
+; KNL_64-LABEL: test_scatter_16f64:
+; KNL_64:       # BB#0:
+; KNL_64-NEXT:    vpmovsxbd %xmm2, %zmm2
+; KNL_64-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; KNL_64-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; KNL_64-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_64-NEXT:    vscatterqpd %zmm3, (,%zmm0) {%k1}
+; KNL_64-NEXT:    vscatterqpd %zmm4, (,%zmm1) {%k2}
+; KNL_64-NEXT:    retq
+;
+; KNL_32-LABEL: test_scatter_16f64:
+; KNL_32:       # BB#0:
+; KNL_32-NEXT:    pushl %ebp
+; KNL_32-NEXT:  .Ltmp9:
+; KNL_32-NEXT:    .cfi_def_cfa_offset 8
+; KNL_32-NEXT:  .Ltmp10:
+; KNL_32-NEXT:    .cfi_offset %ebp, -8
+; KNL_32-NEXT:    movl %esp, %ebp
+; KNL_32-NEXT:  .Ltmp11:
+; KNL_32-NEXT:    .cfi_def_cfa_register %ebp
+; KNL_32-NEXT:    andl $-64, %esp
+; KNL_32-NEXT:    subl $64, %esp
+; KNL_32-NEXT:    vpmovsxbd %xmm1, %zmm1
+; KNL_32-NEXT:    vpandd .LCPI38_0{1to16}, %zmm1, %zmm1
+; KNL_32-NEXT:    vptestmd %zmm1, %zmm1, %k1
+; KNL_32-NEXT:    vmovapd 8(%ebp), %zmm1
+; KNL_32-NEXT:    kshiftrw $8, %k1, %k2
+; KNL_32-NEXT:    vscatterdpd %zmm2, (,%ymm0) {%k1}
+; KNL_32-NEXT:    vextracti64x4 $1, %zmm0, %ymm0
+; KNL_32-NEXT:    vscatterdpd %zmm1, (,%ymm0) {%k2}
+; KNL_32-NEXT:    movl %ebp, %esp
+; KNL_32-NEXT:    popl %ebp
+; KNL_32-NEXT:    retl
+;
+; SKX-LABEL: test_scatter_16f64:
+; SKX:       # BB#0:
+; SKX-NEXT:    vpmovsxbd %xmm2, %zmm2
+; SKX-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm2, %zmm2
+; SKX-NEXT:    vptestmd %zmm2, %zmm2, %k1
+; SKX-NEXT:    kshiftrw $8, %k1, %k2
+; SKX-NEXT:    vscatterqpd %zmm3, (,%zmm0) {%k1}
+; SKX-NEXT:    vscatterqpd %zmm4, (,%zmm1) {%k2}
+; SKX-NEXT:    retq
+  call void @llvm.masked.scatter.v16f64(<16 x double> %src0, <16 x double*> %ptrs, i32 4, <16 x i1> %mask)
+  ret void
+}
+declare void @llvm.masked.scatter.v16f64(<16 x double> %src0, <16 x double*> %ptrs, i32, <16 x i1> %mask)

diff --git a/test/CodeGen/X86/masked_memop.ll b/test/CodeGen/X86/masked_memop.ll
index 1a9cf008e86900a6d0e60442a381f8c160316de2..0662d1b22eda9b2de7389061ab4595133957b17d 100644 (file)
--- a/test/CodeGen/X86/masked_memop.ll
+++ b/test/CodeGen/X86/masked_memop.ll
@@ -1,7 +1,7 @@
-; RUN: llc -mtriple=x86_64-apple-darwin  -mcpu=knl < %s | FileCheck %s -check-prefix=AVX512
-; RUN: llc -mtriple=x86_64-apple-darwin  -mcpu=core-avx2 < %s | FileCheck %s -check-prefix=AVX2
-; RUN: opt -mtriple=x86_64-apple-darwin -codegenprepare -mcpu=corei7-avx -S < %s | FileCheck %s -check-prefix=AVX_SCALAR
-; RUN: llc -mtriple=x86_64-apple-darwin  -mcpu=skx < %s | FileCheck %s -check-prefix=SKX
+; RUN: llc -mtriple=x86_64-apple-darwin  -mcpu=knl < %s | FileCheck %s --check-prefix=AVX512
+; RUN: llc -mtriple=x86_64-apple-darwin  -mcpu=core-avx2 < %s | FileCheck %s --check-prefix=AVX2
+; RUN: opt -mtriple=x86_64-apple-darwin -codegenprepare -mcpu=corei7-avx -S < %s | FileCheck %s --check-prefix=AVX_SCALAR
+; RUN: llc -mtriple=x86_64-apple-darwin  -mcpu=skx < %s | FileCheck %s --check-prefix=SKX
 
 ; AVX512-LABEL: test1
 ; AVX512: vmovdqu32       (%rdi), %zmm0 {%k1} {z}
@@ -274,6 +274,15 @@ define <2 x i32> @test17(<2 x i32> %trigger, <2 x i32>* %addr, <2 x i32> %dst) {
 ; AVX2-NOT: blend
 ; AVX2: ret
 define <2 x float> @test18(<2 x i32> %trigger, <2 x float>* %addr) {
+; SKX-LABEL: test18:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; SKX-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; SKX-NEXT:    vpcmpeqq %xmm1, %xmm0, %k0
+; SKX-NEXT:    kshiftlw $2, %k0, %k0
+; SKX-NEXT:    kshiftrw $2, %k0, %k1
+; SKX-NEXT:    vmovups (%rdi), %xmm0 {%k1} {z}
+; SKX-NEXT:    retq
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer
   %res = call <2 x float> @llvm.masked.load.v2f32(<2 x float>* %addr, i32 4, <2 x i1>%mask, <2 x float>undef)
   ret <2 x float> %res
@@ -363,3 +372,77 @@ define <16 x %mystruct*> @test24(<16 x i1> %mask, <16 x %mystruct*>* %addr) {
   %res = call <16 x %mystruct*> @llvm.masked.load.v16p0mystruct(<16 x %mystruct*>* %addr, i32 4, <16 x i1>%mask, <16 x %mystruct*>zeroinitializer)
   ret <16 x %mystruct*> %res
 }
+
+define void @test_store_16i64(<16 x i64>* %ptrs, <16 x i1> %mask, <16 x i64> %src0)  {
+; SKX-LABEL: test_store_16i64:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vpmovb2m %xmm0, %k1
+; SKX-NEXT:    vmovdqu64 %zmm1, (%rdi) {%k1}
+; SKX-NEXT:    kshiftrw $8, %k1, %k1
+; SKX-NEXT:    vmovdqu64 %zmm2, 64(%rdi) {%k1}
+; SKX-NEXT:    retq
+  call void @llvm.masked.store.v16i64(<16 x i64> %src0, <16 x i64>* %ptrs, i32 4, <16 x i1> %mask)
+  ret void
+}
+declare void @llvm.masked.store.v16i64(<16 x i64> %src0, <16 x i64>* %ptrs, i32, <16 x i1> %mask)
+define void @test_store_16f64(<16 x double>* %ptrs, <16 x i1> %mask, <16 x double> %src0)  {
+; SKX-LABEL: test_store_16f64:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vpmovb2m %xmm0, %k1
+; SKX-NEXT:    vmovupd %zmm1, (%rdi) {%k1}
+; SKX-NEXT:    kshiftrw $8, %k1, %k1
+; SKX-NEXT:    vmovupd %zmm2, 64(%rdi) {%k1}
+; SKX-NEXT:    retq
+  call void @llvm.masked.store.v16f64(<16 x double> %src0, <16 x double>* %ptrs, i32 4, <16 x i1> %mask)
+  ret void
+}
+declare void @llvm.masked.store.v16f64(<16 x double> %src0, <16 x double>* %ptrs, i32, <16 x i1> %mask)
+define <16 x i64> @test_load_16i64(<16 x i64>* %ptrs, <16 x i1> %mask, <16 x i64> %src0)  {
+; SKX-LABEL: test_load_16i64:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vpmovb2m %xmm0, %k1
+; SKX-NEXT:    vmovdqu64 (%rdi), %zmm1 {%k1}
+; SKX-NEXT:    kshiftrw $8, %k1, %k1
+; SKX-NEXT:    vmovdqu64 64(%rdi), %zmm2 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    vmovaps %zmm2, %zmm1
+; SKX-NEXT:    retq
+  %res = call <16 x i64> @llvm.masked.load.v16i64(<16 x i64>* %ptrs, i32 4, <16 x i1> %mask, <16 x i64> %src0)
+  ret <16 x i64> %res
+}
+declare <16 x i64> @llvm.masked.load.v16i64(<16 x i64>* %ptrs, i32, <16 x i1> %mask, <16 x i64> %src0)
+define <16 x double> @test_load_16f64(<16 x double>* %ptrs, <16 x i1> %mask, <16 x double> %src0)  {
+; SKX-LABEL: test_load_16f64:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vpmovb2m %xmm0, %k1
+; SKX-NEXT:    vmovupd (%rdi), %zmm1 {%k1}
+; SKX-NEXT:    kshiftrw $8, %k1, %k1
+; SKX-NEXT:    vmovupd 64(%rdi), %zmm2 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    vmovaps %zmm2, %zmm1
+; SKX-NEXT:    retq
+  %res = call <16 x double> @llvm.masked.load.v16f64(<16 x double>* %ptrs, i32 4, <16 x i1> %mask, <16 x double> %src0)
+  ret <16 x double> %res
+}
+declare <16 x double> @llvm.masked.load.v16f64(<16 x double>* %ptrs, i32, <16 x i1> %mask, <16 x double> %src0)
+
+define <32 x double> @test_load_32f64(<32 x double>* %ptrs, <32 x i1> %mask, <32 x double> %src0)  {
+; SKX-LABEL: test_load_32f64:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vpmovb2m %ymm0, %k1
+; SKX-NEXT:    vmovupd (%rdi), %zmm1 {%k1}
+; SKX-NEXT:    kshiftrd $16, %k1, %k2
+; SKX-NEXT:    vmovupd 128(%rdi), %zmm3 {%k2}
+; SKX-NEXT:    kshiftrw $8, %k1, %k1
+; SKX-NEXT:    vmovupd 64(%rdi), %zmm2 {%k1}
+; SKX-NEXT:    kshiftrw $8, %k2, %k1
+; SKX-NEXT:    vmovupd 192(%rdi), %zmm4 {%k1}
+; SKX-NEXT:    vmovaps %zmm1, %zmm0
+; SKX-NEXT:    vmovaps %zmm2, %zmm1
+; SKX-NEXT:    vmovaps %zmm3, %zmm2
+; SKX-NEXT:    vmovaps %zmm4, %zmm3
+; SKX-NEXT:    retq
+  %res = call <32 x double> @llvm.masked.load.v32f64(<32 x double>* %ptrs, i32 4, <32 x i1> %mask, <32 x double> %src0)
+  ret <32 x double> %res
+}
+declare <32 x double> @llvm.masked.load.v32f64(<32 x double>* %ptrs, i32, <32 x i1> %mask, <32 x double> %src0)