diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -19859,6 +19859,48 @@
   return DAG.getBitcast(VT, DAG.getBuildVector(VecVT, DL, Ops));
 }
 
+// Attempt to merge nested concat_vectors/undefs.
+// Fold concat_vectors(concat_vectors(x,y,z,w),u,u,concat_vectors(a,b,c,d))
+//  --> concat_vectors(x,y,z,w,u,u,u,u,u,u,u,u,a,b,c,d)
+static SDValue combineConcatVectorOfConcatVectors(SDNode *N, SelectionDAG &DAG) {
+  EVT VT = N->getValueType(0);
+
+  // TODO: Add scalable vector type support.
+  if (!VT.isFixedLengthVector())
+      return SDValue();
+
+  // Ensure we're concatenating UNDEF and CONCAT_VECTORS nodes of similar types.
+  EVT SubVT;
+  SDValue FirstConcat;
+  for (const SDValue &Op : N->ops()) {
+    if (Op.isUndef())
+      continue;
+    if (Op.getOpcode() != ISD::CONCAT_VECTORS)
+      return SDValue();
+    if (!FirstConcat) {
+      SubVT = Op.getOperand(0).getValueType();
+      if (!DAG.getTargetLoweringInfo().isTypeLegal(SubVT))
+        return SDValue();
+      FirstConcat = Op;
+      continue;
+    }
+    if (SubVT != Op.getOperand(0).getValueType())
+      return SDValue();
+  }
+  assert(FirstConcat && "Concat of all-undefs found");
+
+  SmallVector<SDValue> ConcatOps;
+  for (const SDValue &Op : N->ops()) {
+    if (Op.isUndef()) {
+      ConcatOps.append(FirstConcat->getNumOperands(), DAG.getUNDEF(SubVT));
+      continue;
+    }
+    for (const SDValue &Sub : Op->ops())
+      ConcatOps.push_back(Sub);
+  }
+  return DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N), VT, ConcatOps);
+}
+
 // Check to see if this is a CONCAT_VECTORS of a bunch of EXTRACT_SUBVECTOR
 // operations. If so, and if the EXTRACT_SUBVECTOR vector inputs come from at
 // most two distinct vectors the same size as the result, attempt to turn this
@@ -20122,9 +20164,13 @@
     return V;
 
   // Fold CONCAT_VECTORS of EXTRACT_SUBVECTOR (or undef) to VECTOR_SHUFFLE.
-  if (Level < AfterLegalizeVectorOps && TLI.isTypeLegal(VT))
+  if (Level < AfterLegalizeVectorOps && TLI.isTypeLegal(VT)) {
+    if (SDValue V = combineConcatVectorOfConcatVectors(N, DAG))
+      return V;
+
     if (SDValue V = combineConcatVectorOfExtracts(N, DAG))
       return V;
+  }
 
   if (SDValue V = combineConcatVectorOfCasts(N, DAG))
     return V;
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -13599,7 +13599,7 @@
   // becomes
   //    (bitconvert (concat_vectors (v4i16 (bitconvert LHS)), RHS))
 
-  if (N1Opc != ISD::BITCAST)
+  if (N1Opc != ISD::BITCAST || N->getNumOperands() != 2)
     return SDValue();
   SDValue RHS = N1->getOperand(0);
   MVT RHSTy = RHS.getValueType().getSimpleVT();
diff --git a/llvm/lib/Target/ARM/ARMISelLowering.cpp b/llvm/lib/Target/ARM/ARMISelLowering.cpp
--- a/llvm/lib/Target/ARM/ARMISelLowering.cpp
+++ b/llvm/lib/Target/ARM/ARMISelLowering.cpp
@@ -8823,54 +8823,68 @@
 
 static SDValue LowerCONCAT_VECTORS_i1(SDValue Op, SelectionDAG &DAG,
                                       const ARMSubtarget *ST) {
-  SDValue V1 = Op.getOperand(0);
-  SDValue V2 = Op.getOperand(1);
   SDLoc dl(Op);
-  EVT VT = Op.getValueType();
-  EVT Op1VT = V1.getValueType();
-  EVT Op2VT = V2.getValueType();
-  unsigned NumElts = VT.getVectorNumElements();
-
-  assert(Op1VT == Op2VT && "Operand types don't match!");
-  assert(VT.getScalarSizeInBits() == 1 &&
+  assert(Op.getValueType().getScalarSizeInBits() == 1 &&
+         "Unexpected custom CONCAT_VECTORS lowering");
+  assert(isPowerOf2_32(Op.getNumOperands()) &&
          "Unexpected custom CONCAT_VECTORS lowering");
   assert(ST->hasMVEIntegerOps() &&
          "CONCAT_VECTORS lowering only supported for MVE");
 
-  SDValue NewV1 = PromoteMVEPredVector(dl, V1, Op1VT, DAG);
-  SDValue NewV2 = PromoteMVEPredVector(dl, V2, Op2VT, DAG);
-
-  // We now have Op1 + Op2 promoted to vectors of integers, where v8i1 gets
-  // promoted to v8i16, etc.
-
-  MVT ElType = getVectorTyFromPredicateVector(VT).getScalarType().getSimpleVT();
-
-  // Extract the vector elements from Op1 and Op2 one by one and truncate them
-  // to be the right size for the destination. For example, if Op1 is v4i1 then
-  // the promoted vector is v4i32. The result of concatentation gives a v8i1,
-  // which when promoted is v8i16. That means each i32 element from Op1 needs
-  // truncating to i16 and inserting in the result.
-  EVT ConcatVT = MVT::getVectorVT(ElType, NumElts);
-  SDValue ConVec = DAG.getNode(ISD::UNDEF, dl, ConcatVT);
-  auto ExractInto = [&DAG, &dl](SDValue NewV, SDValue ConVec, unsigned &j) {
-    EVT NewVT = NewV.getValueType();
-    EVT ConcatVT = ConVec.getValueType();
-    for (unsigned i = 0, e = NewVT.getVectorNumElements(); i < e; i++, j++) {
-      SDValue Elt = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, MVT::i32, NewV,
-                                DAG.getIntPtrConstant(i, dl));
-      ConVec = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, ConcatVT, ConVec, Elt,
-                           DAG.getConstant(j, dl, MVT::i32));
-    }
-    return ConVec;
+  auto ConcatPair = [&](SDValue V1, SDValue V2) {
+    EVT Op1VT = V1.getValueType();
+    EVT Op2VT = V2.getValueType();
+    assert(Op1VT == Op2VT && "Operand types don't match!");
+    EVT VT = Op1VT.getDoubleNumVectorElementsVT(*DAG.getContext());
+
+    SDValue NewV1 = PromoteMVEPredVector(dl, V1, Op1VT, DAG);
+    SDValue NewV2 = PromoteMVEPredVector(dl, V2, Op2VT, DAG);
+
+    // We now have Op1 + Op2 promoted to vectors of integers, where v8i1 gets
+    // promoted to v8i16, etc.
+    MVT ElType =
+        getVectorTyFromPredicateVector(VT).getScalarType().getSimpleVT();
+    unsigned NumElts = 2 * Op1VT.getVectorNumElements();
+
+    // Extract the vector elements from Op1 and Op2 one by one and truncate them
+    // to be the right size for the destination. For example, if Op1 is v4i1
+    // then the promoted vector is v4i32. The result of concatentation gives a
+    // v8i1, which when promoted is v8i16. That means each i32 element from Op1
+    // needs truncating to i16 and inserting in the result.
+    EVT ConcatVT = MVT::getVectorVT(ElType, NumElts);
+    SDValue ConVec = DAG.getNode(ISD::UNDEF, dl, ConcatVT);
+    auto ExractInto = [&DAG, &dl](SDValue NewV, SDValue ConVec, unsigned &j) {
+      EVT NewVT = NewV.getValueType();
+      EVT ConcatVT = ConVec.getValueType();
+      for (unsigned i = 0, e = NewVT.getVectorNumElements(); i < e; i++, j++) {
+        SDValue Elt = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, MVT::i32, NewV,
+                                  DAG.getIntPtrConstant(i, dl));
+        ConVec = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, ConcatVT, ConVec, Elt,
+                             DAG.getConstant(j, dl, MVT::i32));
+      }
+      return ConVec;
+    };
+    unsigned j = 0;
+    ConVec = ExractInto(NewV1, ConVec, j);
+    ConVec = ExractInto(NewV2, ConVec, j);
+
+    // Now return the result of comparing the subvector with zero,
+    // which will generate a real predicate, i.e. v4i1, v8i1 or v16i1.
+    return DAG.getNode(ARMISD::VCMPZ, dl, VT, ConVec,
+                       DAG.getConstant(ARMCC::NE, dl, MVT::i32));
   };
-  unsigned j = 0;
-  ConVec = ExractInto(NewV1, ConVec, j);
-  ConVec = ExractInto(NewV2, ConVec, j);
 
-  // Now return the result of comparing the subvector with zero,
-  // which will generate a real predicate, i.e. v4i1, v8i1 or v16i1.
-  return DAG.getNode(ARMISD::VCMPZ, dl, VT, ConVec,
-                     DAG.getConstant(ARMCC::NE, dl, MVT::i32));
+  // Concat each pair of subvectors and pack into the lower half of the array.
+  SmallVector<SDValue> ConcatOps(Op->op_begin(), Op->op_end());
+  while (ConcatOps.size() > 1) {
+    for (unsigned I = 0, E = ConcatOps.size(); I != E; I += 2) {
+      SDValue V1 = ConcatOps[I];
+      SDValue V2 = ConcatOps[I + 1];
+      ConcatOps[I / 2] = ConcatPair(V1, V2);
+    }
+    ConcatOps.resize(ConcatOps.size() / 2);
+  }
+  return ConcatOps[0];
 }
 
 static SDValue LowerCONCAT_VECTORS(SDValue Op, SelectionDAG &DAG,
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-extload-truncstore.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-extload-truncstore.ll
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-extload-truncstore.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-extload-truncstore.ll
@@ -1290,37 +1290,30 @@
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v27, v26
-; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 4, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v25, 4
-; LMULMAX1-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v27, 0
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v10, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 12, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 8
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v11, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 4
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 16, e8, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v26, 8
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 12
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, m1, ta, mu
-; LMULMAX1-NEXT:    vse8.v v25, (a0)
+; LMULMAX1-NEXT:    vse8.v v26, (a0)
 ; LMULMAX1-NEXT:    ret
 ;
 ; LMULMAX4-LABEL: truncstore_v16i32_v16i8:
@@ -1624,43 +1617,36 @@
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v27, v26
-; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v25, 2
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v27, 0
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v10, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v11, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 2
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v26, 4
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 6
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vse8.v v25, (a0)
+; LMULMAX1-NEXT:    vse8.v v26, (a0)
 ; LMULMAX1-NEXT:    ret
 ;
 ; LMULMAX4-LABEL: truncstore_v8i64_v8i8:
@@ -1685,37 +1671,30 @@
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v27, v26
-; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e16, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v25, 2
-; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e16, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v27, 0
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v10, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e16, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v11, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 2
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v26, 4
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 6
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, m1, ta, mu
-; LMULMAX1-NEXT:    vse16.v v25, (a0)
+; LMULMAX1-NEXT:    vse16.v v26, (a0)
 ; LMULMAX1-NEXT:    ret
 ;
 ; LMULMAX4-LABEL: truncstore_v8i64_v8i16:
@@ -1779,88 +1758,68 @@
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v27, v25
-; LMULMAX1-NEXT:    vslideup.vi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v26, v9, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v26, 2
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v28, v26
-; LMULMAX1-NEXT:    vslideup.vi v28, v27, 0
+; LMULMAX1-NEXT:    vsetivli zero, 4, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v10, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v10, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v29, v25
-; LMULMAX1-NEXT:    vslideup.vi v29, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 4
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v11, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v11, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v27, 2
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v28, v29, 4
-; LMULMAX1-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v27, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v28, 0
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 6
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v12, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v12, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v29, v25
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
+; LMULMAX1-NEXT:    vsetivli zero, 10, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 8
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v13, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v13, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 2
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v29, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
+; LMULMAX1-NEXT:    vsetivli zero, 12, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 10
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v14, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v14, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
+; LMULMAX1-NEXT:    vsetivli zero, 14, e8, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 12
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v15, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v15, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v28, 2
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
+; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 16, e8, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v26, 8
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 14
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, m1, ta, mu
-; LMULMAX1-NEXT:    vse8.v v27, (a0)
+; LMULMAX1-NEXT:    vse8.v v25, (a0)
 ; LMULMAX1-NEXT:    ret
 ;
 ; LMULMAX4-LABEL: truncstore_v16i64_v16i8:
@@ -1897,67 +1856,54 @@
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v28, v25
-; LMULMAX1-NEXT:    vslideup.vi v28, v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v26, v9, 0
-; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v28, v26, 2
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, m1, tu, mu
+; LMULMAX1-NEXT:    vsetivli zero, 2, e16, m1, tu, mu
 ; LMULMAX1-NEXT:    vmv1r.v v27, v26
-; LMULMAX1-NEXT:    vslideup.vi v27, v28, 0
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v10, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v29, v25
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 4, e16, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v11, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 2
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e16, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 4
+; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
+; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v29, 4
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 6
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v12, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v12, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v29, v25
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e16, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v13, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v13, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 2
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e16, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v29, 0
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v14, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v14, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e16, m1, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v15, 0
+; LMULMAX1-NEXT:    vnsrl.wi v25, v15, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v28, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v25, v28, 2
+; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 6
 ; LMULMAX1-NEXT:    addi a1, a0, 16
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, m1, ta, mu
 ; LMULMAX1-NEXT:    vse16.v v26, (a1)
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-conv.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-conv.ll
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-conv.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-conv.ll
@@ -202,15 +202,15 @@
 ;
 ; LMULMAX1-LABEL: fpround_v8f64_v8f16:
 ; LMULMAX1:       # %bb.0:
-; LMULMAX1-NEXT:    addi sp, sp, -48
-; LMULMAX1-NEXT:    .cfi_def_cfa_offset 48
+; LMULMAX1-NEXT:    addi sp, sp, -32
+; LMULMAX1-NEXT:    .cfi_def_cfa_offset 32
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; LMULMAX1-NEXT:    vle64.v v25, (a0)
-; LMULMAX1-NEXT:    addi a2, a0, 32
+; LMULMAX1-NEXT:    addi a2, a0, 16
 ; LMULMAX1-NEXT:    vle64.v v26, (a2)
 ; LMULMAX1-NEXT:    addi a2, a0, 48
 ; LMULMAX1-NEXT:    vle64.v v27, (a2)
-; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    addi a0, a0, 32
 ; LMULMAX1-NEXT:    vle64.v v28, (a0)
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.rod.f.f.w v29, v27
@@ -222,35 +222,25 @@
 ; LMULMAX1-NEXT:    vfncvt.rod.f.f.w v27, v28
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v28, v27
-; LMULMAX1-NEXT:    addi a0, sp, 20
+; LMULMAX1-NEXT:    addi a0, sp, 24
 ; LMULMAX1-NEXT:    vse16.v v28, (a0)
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.rod.f.f.w v27, v26
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v26, v27
-; LMULMAX1-NEXT:    addi a0, sp, 24
-; LMULMAX1-NEXT:    vse16.v v26, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 24
-; LMULMAX1-NEXT:    vle16.v v26, (a0)
-; LMULMAX1-NEXT:    addi a0, sp, 40
+; LMULMAX1-NEXT:    addi a0, sp, 20
 ; LMULMAX1-NEXT:    vse16.v v26, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.rod.f.f.w v26, v25
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v25, v26
 ; LMULMAX1-NEXT:    addi a0, sp, 16
 ; LMULMAX1-NEXT:    vse16.v v25, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 16
-; LMULMAX1-NEXT:    vle16.v v25, (a0)
-; LMULMAX1-NEXT:    addi a0, sp, 32
-; LMULMAX1-NEXT:    vse16.v v25, (a0)
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 32
+; LMULMAX1-NEXT:    addi a0, sp, 16
 ; LMULMAX1-NEXT:    vle16.v v25, (a0)
 ; LMULMAX1-NEXT:    vse16.v v25, (a1)
-; LMULMAX1-NEXT:    addi sp, sp, 48
+; LMULMAX1-NEXT:    addi sp, sp, 32
 ; LMULMAX1-NEXT:    ret
   %a = load <8 x double>, <8 x double>* %x
   %d = fptrunc <8 x double> %a to <8 x half>
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp2i.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp2i.ll
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp2i.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp2i.ll
@@ -456,43 +456,36 @@
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v29, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v29, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v30, v29
-; LMULMAX1-NEXT:    vslideup.vi v30, v27, 0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v29, v27, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v27, v28
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v30, v27, 2
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v27, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v30, 0
+; LMULMAX1-NEXT:    vslideup.vi v29, v27, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v28, v26
+; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v27, v26
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v26, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v27, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v26, 0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v29, v26, 4
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v26, v25
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v25, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v29, 4
+; LMULMAX1-NEXT:    vslideup.vi v29, v25, 6
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vse8.v v27, (a1)
+; LMULMAX1-NEXT:    vse8.v v29, (a1)
 ; LMULMAX1-NEXT:    ret
   %a = load <8 x double>, <8 x double>* %x
   %d = fptosi <8 x double> %a to <8 x i8>
@@ -530,43 +523,36 @@
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v29, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v29, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v30, v29
-; LMULMAX1-NEXT:    vslideup.vi v30, v27, 0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v29, v27, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v27, v28
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v27, v27, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v30, v27, 2
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v27, 0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v30, 0
+; LMULMAX1-NEXT:    vslideup.vi v29, v27, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v28, v26
+; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v27, v26
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
-; LMULMAX1-NEXT:    vnsrl.wi v26, v28, 0
+; LMULMAX1-NEXT:    vnsrl.wi v26, v27, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v26, v26, 0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v26, 0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v29, v26, 4
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v26, v25
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v26, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vnsrl.wi v25, v25, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v29, v25, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v27, v29, 4
+; LMULMAX1-NEXT:    vslideup.vi v29, v25, 6
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vse8.v v27, (a1)
+; LMULMAX1-NEXT:    vse8.v v29, (a1)
 ; LMULMAX1-NEXT:    ret
   %a = load <8 x double>, <8 x double>* %x
   %d = fptoui <8 x double> %a to <8 x i8>
@@ -585,7 +571,7 @@
 ;
 ; LMULMAX1-LABEL: fp2si_v8f64_v8i1:
 ; LMULMAX1:       # %bb.0:
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v25, 0
 ; LMULMAX1-NEXT:    vmclr.m v0
 ; LMULMAX1-NEXT:    vmerge.vim v26, v25, 1, v0
@@ -596,58 +582,43 @@
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v27, 0
 ; LMULMAX1-NEXT:    vmerge.vim v28, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v29, v26
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
-; LMULMAX1-NEXT:    vmerge.vim v28, v25, 1, v0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v28, 0
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
+; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
+; LMULMAX1-NEXT:    vmerge.vim v26, v25, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v29, v9
-; LMULMAX1-NEXT:    vand.vi v29, v29, 1
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vmerge.vim v29, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v28, v29, 2
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v28, v9
+; LMULMAX1-NEXT:    vand.vi v28, v28, 1
 ; LMULMAX1-NEXT:    vmsne.vi v0, v28, 0
-; LMULMAX1-NEXT:    vmerge.vim v28, v25, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v29, 0
-; LMULMAX1-NEXT:    vmclr.m v0
-; LMULMAX1-NEXT:    vmerge.vim v30, v29, 1, v0
+; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
+; LMULMAX1-NEXT:    vmerge.vim v28, v27, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v30, v28, 0
+; LMULMAX1-NEXT:    vslideup.vi v26, v28, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmsne.vi v0, v30, 0
-; LMULMAX1-NEXT:    vmerge.vim v28, v29, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v29, v10
-; LMULMAX1-NEXT:    vand.vi v29, v29, 1
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vmerge.vim v29, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v29, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
 ; LMULMAX1-NEXT:    vmerge.vim v26, v25, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v29, v11
-; LMULMAX1-NEXT:    vand.vi v29, v29, 1
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
+; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v28, v10
+; LMULMAX1-NEXT:    vand.vi v28, v28, 1
+; LMULMAX1-NEXT:    vmsne.vi v0, v28, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vmerge.vim v27, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 2
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vmerge.vim v28, v27, 1, v0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v28, 4
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
 ; LMULMAX1-NEXT:    vmerge.vim v25, v25, 1, v0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; LMULMAX1-NEXT:    vfncvt.rtz.x.f.w v26, v11
+; LMULMAX1-NEXT:    vand.vi v26, v26, 1
+; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
+; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
+; LMULMAX1-NEXT:    vmerge.vim v26, v27, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v28, v25, 4
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 6
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmsne.vi v0, v28, 0
+; LMULMAX1-NEXT:    vmsne.vi v0, v25, 0
 ; LMULMAX1-NEXT:    ret
   %z = fptosi <8 x double> %x to <8 x i1>
   ret <8 x i1> %z
@@ -664,7 +635,7 @@
 ;
 ; LMULMAX1-LABEL: fp2ui_v8f64_v8i1:
 ; LMULMAX1:       # %bb.0:
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v25, 0
 ; LMULMAX1-NEXT:    vmclr.m v0
 ; LMULMAX1-NEXT:    vmerge.vim v26, v25, 1, v0
@@ -675,58 +646,43 @@
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
 ; LMULMAX1-NEXT:    vmv.v.i v27, 0
 ; LMULMAX1-NEXT:    vmerge.vim v28, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vmv1r.v v29, v26
-; LMULMAX1-NEXT:    vslideup.vi v29, v28, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
-; LMULMAX1-NEXT:    vmerge.vim v28, v25, 1, v0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v28, 0
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
+; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
+; LMULMAX1-NEXT:    vmerge.vim v26, v25, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v29, v9
-; LMULMAX1-NEXT:    vand.vi v29, v29, 1
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vmerge.vim v29, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v28, v29, 2
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v28, v9
+; LMULMAX1-NEXT:    vand.vi v28, v28, 1
 ; LMULMAX1-NEXT:    vmsne.vi v0, v28, 0
-; LMULMAX1-NEXT:    vmerge.vim v28, v25, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmv.v.i v29, 0
-; LMULMAX1-NEXT:    vmclr.m v0
-; LMULMAX1-NEXT:    vmerge.vim v30, v29, 1, v0
+; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
+; LMULMAX1-NEXT:    vmerge.vim v28, v27, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v30, v28, 0
+; LMULMAX1-NEXT:    vslideup.vi v26, v28, 2
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmsne.vi v0, v30, 0
-; LMULMAX1-NEXT:    vmerge.vim v28, v29, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v29, v10
-; LMULMAX1-NEXT:    vand.vi v29, v29, 1
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vmerge.vim v29, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 2, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v29, 0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
 ; LMULMAX1-NEXT:    vmerge.vim v26, v25, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
-; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v29, v11
-; LMULMAX1-NEXT:    vand.vi v29, v29, 1
-; LMULMAX1-NEXT:    vmsne.vi v0, v29, 0
+; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v28, v10
+; LMULMAX1-NEXT:    vand.vi v28, v28, 1
+; LMULMAX1-NEXT:    vmsne.vi v0, v28, 0
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
-; LMULMAX1-NEXT:    vmerge.vim v27, v27, 1, v0
-; LMULMAX1-NEXT:    vsetivli zero, 4, e8, mf4, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v26, v27, 2
-; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf4, ta, mu
+; LMULMAX1-NEXT:    vmerge.vim v28, v27, 1, v0
+; LMULMAX1-NEXT:    vsetivli zero, 6, e8, mf2, tu, mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v28, 4
+; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
 ; LMULMAX1-NEXT:    vmerge.vim v25, v25, 1, v0
+; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; LMULMAX1-NEXT:    vfncvt.rtz.xu.f.w v26, v11
+; LMULMAX1-NEXT:    vand.vi v26, v26, 1
+; LMULMAX1-NEXT:    vmsne.vi v0, v26, 0
+; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf8, ta, mu
+; LMULMAX1-NEXT:    vmerge.vim v26, v27, 1, v0
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e8, mf2, tu, mu
-; LMULMAX1-NEXT:    vslideup.vi v28, v25, 4
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 6
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e8, mf2, ta, mu
-; LMULMAX1-NEXT:    vmsne.vi v0, v28, 0
+; LMULMAX1-NEXT:    vmsne.vi v0, v25, 0
 ; LMULMAX1-NEXT:    ret
   %z = fptoui <8 x double> %x to <8 x i1>
   ret <8 x i1> %z
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-i2fp.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-i2fp.ll
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-i2fp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-i2fp.ll
@@ -466,15 +466,15 @@
 ;
 ; LMULMAX1-LABEL: si2fp_v8i64_v8f16:
 ; LMULMAX1:       # %bb.0:
-; LMULMAX1-NEXT:    addi sp, sp, -48
-; LMULMAX1-NEXT:    .cfi_def_cfa_offset 48
+; LMULMAX1-NEXT:    addi sp, sp, -32
+; LMULMAX1-NEXT:    .cfi_def_cfa_offset 32
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; LMULMAX1-NEXT:    vle64.v v25, (a0)
-; LMULMAX1-NEXT:    addi a2, a0, 32
+; LMULMAX1-NEXT:    addi a2, a0, 16
 ; LMULMAX1-NEXT:    vle64.v v26, (a2)
 ; LMULMAX1-NEXT:    addi a2, a0, 48
 ; LMULMAX1-NEXT:    vle64.v v27, (a2)
-; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    addi a0, a0, 32
 ; LMULMAX1-NEXT:    vle64.v v28, (a0)
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.x.w v29, v27
@@ -486,35 +486,25 @@
 ; LMULMAX1-NEXT:    vfncvt.f.x.w v27, v28
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v28, v27
-; LMULMAX1-NEXT:    addi a0, sp, 20
+; LMULMAX1-NEXT:    addi a0, sp, 24
 ; LMULMAX1-NEXT:    vse16.v v28, (a0)
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.x.w v27, v26
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v26, v27
-; LMULMAX1-NEXT:    addi a0, sp, 24
-; LMULMAX1-NEXT:    vse16.v v26, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 24
-; LMULMAX1-NEXT:    vle16.v v26, (a0)
-; LMULMAX1-NEXT:    addi a0, sp, 40
+; LMULMAX1-NEXT:    addi a0, sp, 20
 ; LMULMAX1-NEXT:    vse16.v v26, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.x.w v26, v25
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v25, v26
 ; LMULMAX1-NEXT:    addi a0, sp, 16
 ; LMULMAX1-NEXT:    vse16.v v25, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 16
-; LMULMAX1-NEXT:    vle16.v v25, (a0)
-; LMULMAX1-NEXT:    addi a0, sp, 32
-; LMULMAX1-NEXT:    vse16.v v25, (a0)
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 32
+; LMULMAX1-NEXT:    addi a0, sp, 16
 ; LMULMAX1-NEXT:    vle16.v v25, (a0)
 ; LMULMAX1-NEXT:    vse16.v v25, (a1)
-; LMULMAX1-NEXT:    addi sp, sp, 48
+; LMULMAX1-NEXT:    addi sp, sp, 32
 ; LMULMAX1-NEXT:    ret
   %a = load <8 x i64>, <8 x i64>* %x
   %d = sitofp <8 x i64> %a to <8 x half>
@@ -536,15 +526,15 @@
 ;
 ; LMULMAX1-LABEL: ui2fp_v8i64_v8f16:
 ; LMULMAX1:       # %bb.0:
-; LMULMAX1-NEXT:    addi sp, sp, -48
-; LMULMAX1-NEXT:    .cfi_def_cfa_offset 48
+; LMULMAX1-NEXT:    addi sp, sp, -32
+; LMULMAX1-NEXT:    .cfi_def_cfa_offset 32
 ; LMULMAX1-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; LMULMAX1-NEXT:    vle64.v v25, (a0)
-; LMULMAX1-NEXT:    addi a2, a0, 32
+; LMULMAX1-NEXT:    addi a2, a0, 16
 ; LMULMAX1-NEXT:    vle64.v v26, (a2)
 ; LMULMAX1-NEXT:    addi a2, a0, 48
 ; LMULMAX1-NEXT:    vle64.v v27, (a2)
-; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    addi a0, a0, 32
 ; LMULMAX1-NEXT:    vle64.v v28, (a0)
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.xu.w v29, v27
@@ -556,35 +546,25 @@
 ; LMULMAX1-NEXT:    vfncvt.f.xu.w v27, v28
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v28, v27
-; LMULMAX1-NEXT:    addi a0, sp, 20
+; LMULMAX1-NEXT:    addi a0, sp, 24
 ; LMULMAX1-NEXT:    vse16.v v28, (a0)
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.xu.w v27, v26
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v26, v27
-; LMULMAX1-NEXT:    addi a0, sp, 24
-; LMULMAX1-NEXT:    vse16.v v26, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 24
-; LMULMAX1-NEXT:    vle16.v v26, (a0)
-; LMULMAX1-NEXT:    addi a0, sp, 40
+; LMULMAX1-NEXT:    addi a0, sp, 20
 ; LMULMAX1-NEXT:    vse16.v v26, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; LMULMAX1-NEXT:    vsetvli zero, zero, e32, mf2, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.xu.w v26, v25
 ; LMULMAX1-NEXT:    vsetvli zero, zero, e16, mf4, ta, mu
 ; LMULMAX1-NEXT:    vfncvt.f.f.w v25, v26
 ; LMULMAX1-NEXT:    addi a0, sp, 16
 ; LMULMAX1-NEXT:    vse16.v v25, (a0)
-; LMULMAX1-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 16
-; LMULMAX1-NEXT:    vle16.v v25, (a0)
-; LMULMAX1-NEXT:    addi a0, sp, 32
-; LMULMAX1-NEXT:    vse16.v v25, (a0)
 ; LMULMAX1-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
-; LMULMAX1-NEXT:    addi a0, sp, 32
+; LMULMAX1-NEXT:    addi a0, sp, 16
 ; LMULMAX1-NEXT:    vle16.v v25, (a0)
 ; LMULMAX1-NEXT:    vse16.v v25, (a1)
-; LMULMAX1-NEXT:    addi sp, sp, 48
+; LMULMAX1-NEXT:    addi sp, sp, 32
 ; LMULMAX1-NEXT:    ret
   %a = load <8 x i64>, <8 x i64>* %x
   %d = uitofp <8 x i64> %a to <8 x half>
diff --git a/llvm/test/CodeGen/X86/merge-consecutive-loads-512.ll b/llvm/test/CodeGen/X86/merge-consecutive-loads-512.ll
--- a/llvm/test/CodeGen/X86/merge-consecutive-loads-512.ll
+++ b/llvm/test/CodeGen/X86/merge-consecutive-loads-512.ll
@@ -8,17 +8,13 @@
 define <8 x double> @merge_8f64_2f64_12u4(<2 x double>* %ptr) nounwind uwtable noinline ssp {
 ; ALL-LABEL: merge_8f64_2f64_12u4:
 ; ALL:       # %bb.0:
-; ALL-NEXT:    vmovups 16(%rdi), %ymm0
-; ALL-NEXT:    vinsertf128 $1, 64(%rdi), %ymm0, %ymm1
-; ALL-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; ALL-NEXT:    vmovups 16(%rdi), %zmm0
 ; ALL-NEXT:    retq
 ;
 ; X86-AVX512F-LABEL: merge_8f64_2f64_12u4:
 ; X86-AVX512F:       # %bb.0:
 ; X86-AVX512F-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-AVX512F-NEXT:    vmovups 16(%eax), %ymm0
-; X86-AVX512F-NEXT:    vinsertf128 $1, 64(%eax), %ymm0, %ymm1
-; X86-AVX512F-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; X86-AVX512F-NEXT:    vmovups 16(%eax), %zmm0
 ; X86-AVX512F-NEXT:    retl
   %ptr0 = getelementptr inbounds <2 x double>, <2 x double>* %ptr, i64 1
   %ptr1 = getelementptr inbounds <2 x double>, <2 x double>* %ptr, i64 2
@@ -35,19 +31,15 @@
 define <8 x double> @merge_8f64_2f64_23z5(<2 x double>* %ptr) nounwind uwtable noinline ssp {
 ; ALL-LABEL: merge_8f64_2f64_23z5:
 ; ALL:       # %bb.0:
-; ALL-NEXT:    vmovups 32(%rdi), %ymm0
-; ALL-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; ALL-NEXT:    vinsertf128 $1, 80(%rdi), %ymm1, %ymm1
-; ALL-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; ALL-NEXT:    vmovdqu64 32(%rdi), %zmm0
+; ALL-NEXT:    vpandq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
 ; ALL-NEXT:    retq
 ;
 ; X86-AVX512F-LABEL: merge_8f64_2f64_23z5:
 ; X86-AVX512F:       # %bb.0:
 ; X86-AVX512F-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-AVX512F-NEXT:    vmovups 32(%eax), %ymm0
-; X86-AVX512F-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; X86-AVX512F-NEXT:    vinsertf128 $1, 80(%eax), %ymm1, %ymm1
-; X86-AVX512F-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; X86-AVX512F-NEXT:    vmovdqu64 32(%eax), %zmm0
+; X86-AVX512F-NEXT:    vpandq {{\.?LCPI[0-9]+_[0-9]+}}, %zmm0, %zmm0
 ; X86-AVX512F-NEXT:    retl
   %ptr0 = getelementptr inbounds <2 x double>, <2 x double>* %ptr, i64 2
   %ptr1 = getelementptr inbounds <2 x double>, <2 x double>* %ptr, i64 3
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll b/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll
--- a/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll
@@ -575,9 +575,7 @@
 define <16 x float> @insert_sub2_4(<16 x float> %base, <4 x float> %sub1, <4 x float> %sub2, <4 x float> %sub3, <4 x float> %sub4) {
 ; ALL-LABEL: insert_sub2_4:
 ; ALL:       # %bb.0:
-; ALL-NEXT:    vinsertf32x4 $2, %xmm3, %zmm0, %zmm1
-; ALL-NEXT:    vmovapd {{.*#+}} zmm2 = [0,1,12,13,4,5,6,7]
-; ALL-NEXT:    vpermt2pd %zmm1, %zmm2, %zmm0
+; ALL-NEXT:    vinsertf32x4 $1, %xmm3, %zmm0, %zmm0
 ; ALL-NEXT:    retq
   %sub12 = shufflevector <4 x float> %sub1, <4 x float> %sub2, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %sub34 = shufflevector <4 x float> %sub3, <4 x float> %sub4, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
diff --git a/llvm/test/CodeGen/X86/x86-interleaved-access.ll b/llvm/test/CodeGen/X86/x86-interleaved-access.ll
--- a/llvm/test/CodeGen/X86/x86-interleaved-access.ll
+++ b/llvm/test/CodeGen/X86/x86-interleaved-access.ll
@@ -369,8 +369,8 @@
 ; AVX512-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm4[4],xmm1[4],xmm4[5],xmm1[5],xmm4[6],xmm1[6],xmm4[7],xmm1[7]
 ; AVX512-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
 ; AVX512-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; AVX512-NEXT:    vinserti128 $1, %xmm1, %ymm3, %ymm1
 ; AVX512-NEXT:    vinserti128 $1, %xmm0, %ymm4, %ymm0
+; AVX512-NEXT:    vinserti128 $1, %xmm1, %ymm3, %ymm1
 ; AVX512-NEXT:    vinserti64x4 $1, %ymm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vmovdqa64 %zmm0, (%rdi)
 ; AVX512-NEXT:    vzeroupper
@@ -1467,15 +1467,15 @@
 ; AVX512-NEXT:    vmovdqu 96(%rdi), %xmm3
 ; AVX512-NEXT:    vmovdqu 112(%rdi), %xmm4
 ; AVX512-NEXT:    vmovdqu 128(%rdi), %xmm5
+; AVX512-NEXT:    vinserti128 $1, 144(%rdi), %ymm3, %ymm3
 ; AVX512-NEXT:    vinserti128 $1, 48(%rdi), %ymm0, %ymm0
+; AVX512-NEXT:    vinserti64x4 $1, %ymm3, %zmm0, %zmm0
+; AVX512-NEXT:    vinserti128 $1, 160(%rdi), %ymm4, %ymm3
 ; AVX512-NEXT:    vinserti128 $1, 64(%rdi), %ymm1, %ymm1
+; AVX512-NEXT:    vinserti64x4 $1, %ymm3, %zmm1, %zmm1
+; AVX512-NEXT:    vinserti128 $1, 176(%rdi), %ymm5, %ymm3
 ; AVX512-NEXT:    vinserti128 $1, 80(%rdi), %ymm2, %ymm2
-; AVX512-NEXT:    vinserti128 $1, 144(%rdi), %ymm3, %ymm3
-; AVX512-NEXT:    vinserti128 $1, 160(%rdi), %ymm4, %ymm4
-; AVX512-NEXT:    vinserti128 $1, 176(%rdi), %ymm5, %ymm5
-; AVX512-NEXT:    vinserti64x4 $1, %ymm3, %zmm0, %zmm0
-; AVX512-NEXT:    vinserti64x4 $1, %ymm4, %zmm1, %zmm1
-; AVX512-NEXT:    vinserti64x4 $1, %ymm5, %zmm2, %zmm2
+; AVX512-NEXT:    vinserti64x4 $1, %ymm3, %zmm2, %zmm2
 ; AVX512-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13]
 ; AVX512-NEXT:    vpshufb %zmm3, %zmm0, %zmm0
 ; AVX512-NEXT:    vpshufb %zmm3, %zmm1, %zmm1