diff --git a/llvm/lib/Target/RISCV/CMakeLists.txt b/llvm/lib/Target/RISCV/CMakeLists.txt
--- a/llvm/lib/Target/RISCV/CMakeLists.txt
+++ b/llvm/lib/Target/RISCV/CMakeLists.txt
@@ -21,10 +21,10 @@
 add_llvm_target(RISCVCodeGen
   RISCVAsmPrinter.cpp
   RISCVCallLowering.cpp
-  RISCVCleanupVSETVLI.cpp
   RISCVExpandAtomicPseudoInsts.cpp
   RISCVExpandPseudoInsts.cpp
   RISCVFrameLowering.cpp
+  RISCVInsertVSETVLI.cpp
   RISCVInstrInfo.cpp
   RISCVInstructionSelector.cpp
   RISCVISelDAGToDAG.cpp
diff --git a/llvm/lib/Target/RISCV/RISCV.h b/llvm/lib/Target/RISCV/RISCV.h
--- a/llvm/lib/Target/RISCV/RISCV.h
+++ b/llvm/lib/Target/RISCV/RISCV.h
@@ -46,8 +46,8 @@
 FunctionPass *createRISCVExpandAtomicPseudoPass();
 void initializeRISCVExpandAtomicPseudoPass(PassRegistry &);
 
-FunctionPass *createRISCVCleanupVSETVLIPass();
-void initializeRISCVCleanupVSETVLIPass(PassRegistry &);
+FunctionPass *createRISCVInsertVSETVLIPass();
+void initializeRISCVInsertVSETVLIPass(PassRegistry &);
 
 InstructionSelector *createRISCVInstructionSelector(const RISCVTargetMachine &,
                                                     RISCVSubtarget &,
diff --git a/llvm/lib/Target/RISCV/RISCVCleanupVSETVLI.cpp b/llvm/lib/Target/RISCV/RISCVCleanupVSETVLI.cpp
deleted file mode 100644
--- a/llvm/lib/Target/RISCV/RISCVCleanupVSETVLI.cpp
+++ /dev/null
@@ -1,163 +0,0 @@
-//===- RISCVCleanupVSETVLI.cpp - Cleanup unneeded VSETVLI instructions ----===//
-//
-// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-//
-//===----------------------------------------------------------------------===//
-//
-// This file implements a function pass that removes duplicate vsetvli
-// instructions within a basic block.
-//
-//===----------------------------------------------------------------------===//
-
-#include "RISCV.h"
-#include "RISCVSubtarget.h"
-#include "llvm/CodeGen/MachineFunctionPass.h"
-using namespace llvm;
-
-#define DEBUG_TYPE "riscv-cleanup-vsetvli"
-#define RISCV_CLEANUP_VSETVLI_NAME "RISCV Cleanup VSETVLI pass"
-
-namespace {
-
-class RISCVCleanupVSETVLI : public MachineFunctionPass {
-public:
-  static char ID;
-
-  RISCVCleanupVSETVLI() : MachineFunctionPass(ID) {
-    initializeRISCVCleanupVSETVLIPass(*PassRegistry::getPassRegistry());
-  }
-  bool runOnMachineFunction(MachineFunction &MF) override;
-  bool runOnMachineBasicBlock(MachineBasicBlock &MBB);
-
-  MachineFunctionProperties getRequiredProperties() const override {
-    return MachineFunctionProperties().set(
-        MachineFunctionProperties::Property::IsSSA);
-  }
-
-  // This pass modifies the program, but does not modify the CFG
-  void getAnalysisUsage(AnalysisUsage &AU) const override {
-    AU.setPreservesCFG();
-    MachineFunctionPass::getAnalysisUsage(AU);
-  }
-
-  StringRef getPassName() const override { return RISCV_CLEANUP_VSETVLI_NAME; }
-};
-
-} // end anonymous namespace
-
-char RISCVCleanupVSETVLI::ID = 0;
-
-INITIALIZE_PASS(RISCVCleanupVSETVLI, DEBUG_TYPE,
-                RISCV_CLEANUP_VSETVLI_NAME, false, false)
-
-static bool isRedundantVSETVLI(MachineInstr &MI, MachineInstr *PrevVSETVLI) {
-  // If we don't have a previous VSET{I}VLI or the VL output isn't dead, we
-  // can't remove this VSETVLI.
-  if (!PrevVSETVLI || !MI.getOperand(0).isDead())
-    return false;
-
-  // Does this VSET{I}VLI use the same VTYPE immediate.
-  int64_t PrevVTYPEImm = PrevVSETVLI->getOperand(2).getImm();
-  int64_t VTYPEImm = MI.getOperand(2).getImm();
-  if (PrevVTYPEImm != VTYPEImm)
-    return false;
-
-  if (MI.getOpcode() == RISCV::PseudoVSETIVLI) {
-    // If the previous opcode wasn't vsetivli we can't compare them.
-    if (PrevVSETVLI->getOpcode() != RISCV::PseudoVSETIVLI)
-      return false;
-
-    // For VSETIVLI, we can just compare the immediates.
-    return PrevVSETVLI->getOperand(1).getImm() == MI.getOperand(1).getImm();
-  }
-
-  assert(MI.getOpcode() == RISCV::PseudoVSETVLI);
-  Register AVLReg = MI.getOperand(1).getReg();
-  Register PrevOutVL = PrevVSETVLI->getOperand(0).getReg();
-
-  // If this VSETVLI isn't changing VL, it is redundant.
-  if (AVLReg == RISCV::X0 && MI.getOperand(0).getReg() == RISCV::X0)
-    return true;
-
-  // If the previous VSET{I}VLI's output (which isn't X0) is fed into this
-  // VSETVLI, this one isn't changing VL so is redundant.
-  // Only perform this on virtual registers to avoid the complexity of having
-  // to work out if the physical register was clobbered somewhere in between.
-  if (AVLReg.isVirtual() && AVLReg == PrevOutVL)
-    return true;
-
-  // If the previous opcode isn't vsetvli we can't do any more comparison.
-  if (PrevVSETVLI->getOpcode() != RISCV::PseudoVSETVLI)
-    return false;
-
-  // Does this VSETVLI use the same AVL register?
-  if (AVLReg != PrevVSETVLI->getOperand(1).getReg())
-    return false;
-
-  // If the AVLReg is X0 we must be setting VL to VLMAX. Keeping VL unchanged
-  // was handled above.
-  if (AVLReg == RISCV::X0) {
-    // This instruction is setting VL to VLMAX, this is redundant if the
-    // previous VSETVLI was also setting VL to VLMAX. But it is not redundant
-    // if they were setting it to any other value or leaving VL unchanged.
-    return PrevOutVL != RISCV::X0;
-  }
-
-  // This vsetvli is redundant.
-  return true;
-}
-
-bool RISCVCleanupVSETVLI::runOnMachineBasicBlock(MachineBasicBlock &MBB) {
-  bool Changed = false;
-  MachineInstr *PrevVSETVLI = nullptr;
-
-  for (auto MII = MBB.begin(), MIE = MBB.end(); MII != MIE;) {
-    MachineInstr &MI = *MII++;
-
-    if (MI.getOpcode() != RISCV::PseudoVSETVLI &&
-        MI.getOpcode() != RISCV::PseudoVSETIVLI) {
-      if (PrevVSETVLI &&
-          (MI.isCall() || MI.modifiesRegister(RISCV::VL) ||
-           MI.modifiesRegister(RISCV::VTYPE))) {
-        // Old VL/VTYPE is overwritten.
-        PrevVSETVLI = nullptr;
-      }
-      continue;
-    }
-
-    if (isRedundantVSETVLI(MI, PrevVSETVLI)) {
-      // This VSETVLI is redundant, remove it.
-      MI.eraseFromParent();
-      Changed = true;
-    } else {
-      // Otherwise update VSET{I}VLI for the next iteration.
-      PrevVSETVLI = &MI;
-    }
-  }
-
-  return Changed;
-}
-
-bool RISCVCleanupVSETVLI::runOnMachineFunction(MachineFunction &MF) {
-  if (skipFunction(MF.getFunction()))
-    return false;
-
-  // Skip if the vector extension is not enabled.
-  const RISCVSubtarget &ST = MF.getSubtarget<RISCVSubtarget>();
-  if (!ST.hasStdExtV())
-    return false;
-
-  bool Changed = false;
-
-  for (MachineBasicBlock &MBB : MF)
-    Changed |= runOnMachineBasicBlock(MBB);
-
-  return Changed;
-}
-
-/// Returns an instance of the Cleanup VSETVLI pass.
-FunctionPass *llvm::createRISCVCleanupVSETVLIPass() {
-  return new RISCVCleanupVSETVLI();
-}
diff --git a/llvm/lib/Target/RISCV/RISCVExpandPseudoInsts.cpp b/llvm/lib/Target/RISCV/RISCVExpandPseudoInsts.cpp
--- a/llvm/lib/Target/RISCV/RISCVExpandPseudoInsts.cpp
+++ b/llvm/lib/Target/RISCV/RISCVExpandPseudoInsts.cpp
@@ -240,7 +240,8 @@
 
 bool RISCVExpandPseudo::expandVSetVL(MachineBasicBlock &MBB,
                                      MachineBasicBlock::iterator MBBI) {
-  assert(MBBI->getNumOperands() == 5 && "Unexpected instruction format");
+  assert(MBBI->getNumExplicitOperands() == 3 && MBBI->getNumOperands() >= 5 &&
+         "Unexpected instruction format");
 
   DebugLoc DL = MBBI->getDebugLoc();
 
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -6436,107 +6436,9 @@
   return TailMBB;
 }
 
-static MachineInstr *elideCopies(MachineInstr *MI,
-                                 const MachineRegisterInfo &MRI) {
-  while (true) {
-    if (!MI->isFullCopy())
-      return MI;
-    if (!Register::isVirtualRegister(MI->getOperand(1).getReg()))
-      return nullptr;
-    MI = MRI.getVRegDef(MI->getOperand(1).getReg());
-    if (!MI)
-      return nullptr;
-  }
-}
-
-static MachineBasicBlock *addVSetVL(MachineInstr &MI, MachineBasicBlock *BB,
-                                    int VLIndex, unsigned SEWIndex,
-                                    RISCVVLMUL VLMul, bool ForceTailAgnostic) {
-  MachineFunction &MF = *BB->getParent();
-  DebugLoc DL = MI.getDebugLoc();
-  const TargetInstrInfo &TII = *MF.getSubtarget().getInstrInfo();
-
-  unsigned Log2SEW = MI.getOperand(SEWIndex).getImm();
-  assert(RISCVVType::isValidSEW(1 << Log2SEW) && "Unexpected SEW");
-  RISCVVSEW ElementWidth = static_cast<RISCVVSEW>(Log2SEW - 3);
-
-  MachineRegisterInfo &MRI = MF.getRegInfo();
-
-  auto BuildVSETVLI = [&]() {
-    if (VLIndex >= 0) {
-      Register DestReg = MRI.createVirtualRegister(&RISCV::GPRRegClass);
-      const MachineOperand &VLOp = MI.getOperand(VLIndex);
-
-      // VL can be a register or an immediate.
-      if (VLOp.isImm())
-        return BuildMI(*BB, MI, DL, TII.get(RISCV::PseudoVSETIVLI))
-            .addReg(DestReg, RegState::Define | RegState::Dead)
-            .addImm(VLOp.getImm());
-
-      Register VLReg = MI.getOperand(VLIndex).getReg();
-      return BuildMI(*BB, MI, DL, TII.get(RISCV::PseudoVSETVLI))
-          .addReg(DestReg, RegState::Define | RegState::Dead)
-          .addReg(VLReg);
-    }
-
-    // With no VL operator in the pseudo, do not modify VL (rd = X0, rs1 = X0).
-    return BuildMI(*BB, MI, DL, TII.get(RISCV::PseudoVSETVLI))
-        .addReg(RISCV::X0, RegState::Define | RegState::Dead)
-        .addReg(RISCV::X0, RegState::Kill);
-  };
-
-  MachineInstrBuilder MIB = BuildVSETVLI();
-
-  // Default to tail agnostic unless the destination is tied to a source. In
-  // that case the user would have some control over the tail values. The tail
-  // policy is also ignored on instructions that only update element 0 like
-  // vmv.s.x or reductions so use agnostic there to match the common case.
-  // FIXME: This is conservatively correct, but we might want to detect that
-  // the input is undefined.
-  bool TailAgnostic = true;
-  unsigned UseOpIdx;
-  if (!ForceTailAgnostic && MI.isRegTiedToUseOperand(0, &UseOpIdx)) {
-    TailAgnostic = false;
-    // If the tied operand is an IMPLICIT_DEF we can keep TailAgnostic.
-    const MachineOperand &UseMO = MI.getOperand(UseOpIdx);
-    MachineInstr *UseMI = MRI.getVRegDef(UseMO.getReg());
-    if (UseMI) {
-      UseMI = elideCopies(UseMI, MRI);
-      if (UseMI && UseMI->isImplicitDef())
-        TailAgnostic = true;
-    }
-  }
-
-  // For simplicity we reuse the vtype representation here.
-  MIB.addImm(RISCVVType::encodeVTYPE(VLMul, ElementWidth,
-                                     /*TailAgnostic*/ TailAgnostic,
-                                     /*MaskAgnostic*/ false));
-
-  // Remove (now) redundant operands from pseudo
-  if (VLIndex >= 0 && MI.getOperand(VLIndex).isReg()) {
-    MI.getOperand(VLIndex).setReg(RISCV::NoRegister);
-    MI.getOperand(VLIndex).setIsKill(false);
-  }
-
-  return BB;
-}
-
 MachineBasicBlock *
 RISCVTargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI,
                                                  MachineBasicBlock *BB) const {
-  uint64_t TSFlags = MI.getDesc().TSFlags;
-
-  if (TSFlags & RISCVII::HasSEWOpMask) {
-    unsigned NumOperands = MI.getNumExplicitOperands();
-    int VLIndex = (TSFlags & RISCVII::HasVLOpMask) ? NumOperands - 2 : -1;
-    unsigned SEWIndex = NumOperands - 1;
-    bool ForceTailAgnostic = TSFlags & RISCVII::ForceTailAgnosticMask;
-
-    RISCVVLMUL VLMul = static_cast<RISCVVLMUL>((TSFlags & RISCVII::VLMulMask) >>
-                                               RISCVII::VLMulShift);
-    return addVSetVL(MI, BB, VLIndex, SEWIndex, VLMul, ForceTailAgnostic);
-  }
-
   switch (MI.getOpcode()) {
   default:
     llvm_unreachable("Unexpected instr type to insert");
diff --git a/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
new file mode 100644
--- /dev/null
+++ b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
@@ -0,0 +1,260 @@
+//===- RISCVInsertVSETVLI.cpp - Insert VSETVLI instructions ---------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file implements a function pass that inserts VSETVLI instructions where
+// needed.
+//
+//===----------------------------------------------------------------------===//
+
+#include "RISCV.h"
+#include "RISCVSubtarget.h"
+#include "llvm/CodeGen/LiveIntervals.h"
+#include "llvm/CodeGen/MachineFunctionPass.h"
+using namespace llvm;
+
+#define DEBUG_TYPE "riscv-insert-vsetvli"
+#define RISCV_INSERT_VSETVLI_NAME "RISCV Insert VSETVLI pass"
+
+namespace {
+
+class RISCVInsertVSETVLI : public MachineFunctionPass {
+  const TargetInstrInfo *TII;
+  MachineRegisterInfo *MRI;
+
+public:
+  static char ID;
+
+  RISCVInsertVSETVLI() : MachineFunctionPass(ID) {
+    initializeRISCVInsertVSETVLIPass(*PassRegistry::getPassRegistry());
+  }
+  bool runOnMachineFunction(MachineFunction &MF) override;
+  bool runOnMachineBasicBlock(MachineBasicBlock &MBB);
+
+  void getAnalysisUsage(AnalysisUsage &AU) const override {
+    AU.setPreservesCFG();
+    MachineFunctionPass::getAnalysisUsage(AU);
+  }
+
+  StringRef getPassName() const override { return RISCV_INSERT_VSETVLI_NAME; }
+
+private:
+  bool maybeInsertVSETVLI(MachineBasicBlock &MBB,
+                          MachineBasicBlock::iterator MII,
+                          const MachineInstr *&PrevVSETVLI);
+};
+
+} // end anonymous namespace
+
+char RISCVInsertVSETVLI::ID = 0;
+
+INITIALIZE_PASS(RISCVInsertVSETVLI, DEBUG_TYPE, RISCV_INSERT_VSETVLI_NAME,
+                false, false)
+
+static MachineInstr *elideCopies(MachineInstr *MI,
+                                 const MachineRegisterInfo *MRI) {
+  while (true) {
+    if (!MI->isFullCopy())
+      return MI;
+    if (!Register::isVirtualRegister(MI->getOperand(1).getReg()))
+      return nullptr;
+    MI = MRI->getVRegDef(MI->getOperand(1).getReg());
+    if (!MI)
+      return nullptr;
+  }
+}
+
+static unsigned computeVType(const MachineInstr &MI, uint64_t TSFlags,
+                             const MachineRegisterInfo *MRI) {
+  unsigned NumOperands = MI.getNumExplicitOperands();
+
+  RISCVVLMUL VLMul = static_cast<RISCVVLMUL>((TSFlags & RISCVII::VLMulMask) >>
+                                             RISCVII::VLMulShift);
+
+  unsigned Log2SEW = MI.getOperand(NumOperands - 1).getImm();
+  assert(RISCVVType::isValidSEW(1 << Log2SEW) && "Unexpected SEW");
+  RISCVVSEW ElementWidth = static_cast<RISCVVSEW>(Log2SEW - 3);
+
+  // Default to tail agnostic unless the destination is tied to a source.
+  // Unless the source is undef. In that case the user would have some control
+  // over the tail values. The tail policy is also ignored on instructions
+  // that only update element 0 like vmv.s.x or reductions so use agnostic
+  // there to match the common case.
+  // FIXME: This is conservatively correct, but we might want to detect that
+  // the input is undefined.
+  bool ForceTailAgnostic = TSFlags & RISCVII::ForceTailAgnosticMask;
+  bool TailAgnostic = true;
+  unsigned UseOpIdx;
+  if (!ForceTailAgnostic && MI.isRegTiedToUseOperand(0, &UseOpIdx)) {
+    TailAgnostic = false;
+    // If the tied operand is an IMPLICIT_DEF we can keep TailAgnostic.
+    const MachineOperand &UseMO = MI.getOperand(UseOpIdx);
+    MachineInstr *UseMI = MRI->getVRegDef(UseMO.getReg());
+    if (UseMI) {
+      UseMI = elideCopies(UseMI, MRI);
+      if (UseMI && UseMI->isImplicitDef())
+        TailAgnostic = true;
+    }
+  }
+
+  return RISCVVType::encodeVTYPE(VLMul, ElementWidth,
+                                 /*TailAgnostic*/ TailAgnostic,
+                                 /*MaskAgnostic*/ false);
+}
+
+bool RISCVInsertVSETVLI::maybeInsertVSETVLI(MachineBasicBlock &MBB,
+                                            MachineBasicBlock::iterator MII,
+                                            const MachineInstr *&PrevVSETVLI) {
+  MachineInstr &MI = *MII;
+
+  uint64_t TSFlags = MI.getDesc().TSFlags;
+  if (!(TSFlags & RISCVII::HasSEWOpMask))
+    return false;
+
+  // Add VL/VTYPE implicit uses.
+  if (TSFlags & RISCVII::HasVLOpMask) {
+    MI.addOperand(
+        MachineOperand::CreateReg(RISCV::VL, /*isDef*/ false, /*isImp*/ true));
+  }
+  MI.addOperand(
+      MachineOperand::CreateReg(RISCV::VTYPE, /*isDef*/ false, /*isImp*/ true));
+
+  unsigned VTypeImm = computeVType(MI, TSFlags, MRI);
+
+  // Invalidate PrevVSETVLI if the vtype doesn't match.
+  if (PrevVSETVLI && PrevVSETVLI->getOperand(2).getImm() != VTypeImm)
+    PrevVSETVLI = nullptr;
+
+  DebugLoc DL = MI.getDebugLoc();
+
+  if (!(TSFlags & RISCVII::HasVLOpMask)) {
+    // This instruction isn't changing VL, if it has the same VTYPE as the last
+    // VSETVLI we're done.
+    if (PrevVSETVLI)
+      return false;
+
+    auto MIB = BuildMI(MBB, MII, DL, TII->get(RISCV::PseudoVSETVLI))
+                   .addReg(RISCV::X0, RegState::Define | RegState::Dead)
+                   .addReg(RISCV::X0, RegState::Kill)
+                   .addImm(VTypeImm)
+                   .addReg(RISCV::VL, RegState::Implicit);
+    PrevVSETVLI = &*MIB;
+    return true;
+  }
+
+  MachineOperand &VLOp = MI.getOperand(MI.getNumExplicitOperands() - 2);
+
+  // First handle the VSETIVLI case.
+  if (VLOp.isImm()) {
+    int64_t Imm = VLOp.getImm();
+
+    // If we have a previous VSETIVLI, we don't need to insert one. The VTYPE
+    // was checked earlier, so just compare the immediate.
+    if (PrevVSETVLI && PrevVSETVLI->getOpcode() == RISCV::PseudoVSETIVLI &&
+        PrevVSETVLI->getOperand(1).getImm() == Imm)
+      return false;
+
+    // TODO: Use X0 as the destination.
+    Register DestReg = MRI->createVirtualRegister(&RISCV::GPRRegClass);
+    auto MIB = BuildMI(MBB, MII, DL, TII->get(RISCV::PseudoVSETIVLI))
+                   .addReg(DestReg, RegState::Define | RegState::Dead)
+                   .addImm(Imm)
+                   .addImm(VTypeImm);
+    PrevVSETVLI = &*MIB;
+
+    return true;
+  }
+
+  // Otherwise we expect VL to be a GPR.
+  Register AVLReg = VLOp.getReg();
+  assert(AVLReg != RISCV::NoRegister && "Unexpected AVL register");
+
+  auto needVSETVLI = [](const MachineInstr *PrevVSETVLI, Register AVLReg) {
+    if (!PrevVSETVLI)
+      return true;
+
+    Register PrevOutVL = PrevVSETVLI->getOperand(0).getReg();
+    // If the previous VSET{I}VLI's output (which isn't X0) is fed into this
+    // VSETVLI, this one isn't changing VL so is redundant.
+    // Only perform this on virtual registers to avoid the complexity of having
+    // to work out if the physical register was clobbered somewhere in between.
+    if (AVLReg.isVirtual() && AVLReg == PrevOutVL)
+      return false;
+
+    if (PrevVSETVLI->getOpcode() == RISCV::PseudoVSETVLI &&
+        AVLReg == PrevVSETVLI->getOperand(1).getReg() &&
+        (AVLReg != RISCV::X0 || PrevOutVL != RISCV::X0))
+      return false;
+
+    return true;
+  };
+
+  if (needVSETVLI(PrevVSETVLI, AVLReg)) {
+    // TODO: Use X0 as the destination if AVLReg is not X0.
+    Register DestReg = MRI->createVirtualRegister(&RISCV::GPRRegClass);
+    auto MIB = BuildMI(MBB, MII, DL, TII->get(RISCV::PseudoVSETVLI))
+                   .addReg(DestReg, RegState::Define | RegState::Dead)
+                   .addReg(AVLReg)
+                   .addImm(VTypeImm);
+    PrevVSETVLI = &*MIB;
+  }
+
+  // Clear the GPR, we don't need it anymore
+  VLOp.setReg(RISCV::NoRegister);
+  VLOp.setIsKill(false);
+
+  return true;
+}
+
+bool RISCVInsertVSETVLI::runOnMachineBasicBlock(MachineBasicBlock &MBB) {
+  bool Changed = false;
+  const MachineInstr *PrevVSETVLI = nullptr;
+
+  for (auto MII = MBB.begin(), MIE = MBB.end(); MII != MIE; ++MII) {
+    MachineInstr &MI = *MII;
+
+    // If this is an explicit VSETVLI/VSETIVILI instruction, cache it as our
+    // previous VL/VTYPE update.
+    if (MI.getOpcode() == RISCV::PseudoVSETVLI ||
+        MI.getOpcode() == RISCV::PseudoVSETIVLI) {
+      PrevVSETVLI = &MI;
+      continue;
+    }
+
+    Changed |= maybeInsertVSETVLI(MBB, MII, PrevVSETVLI);
+
+    if (PrevVSETVLI && (MI.isCall() || MI.modifiesRegister(RISCV::VL) ||
+                        MI.modifiesRegister(RISCV::VTYPE))) {
+      // Old VL/VTYPE is overwritten.
+      PrevVSETVLI = nullptr;
+    }
+  }
+
+  return Changed;
+}
+
+bool RISCVInsertVSETVLI::runOnMachineFunction(MachineFunction &MF) {
+  // Skip if the vector extension is not enabled.
+  const RISCVSubtarget &ST = MF.getSubtarget<RISCVSubtarget>();
+  if (!ST.hasStdExtV())
+    return false;
+
+  const RISCVSubtarget &Subtarget = MF.getSubtarget<RISCVSubtarget>();
+  TII = Subtarget.getInstrInfo();
+  MRI = &MF.getRegInfo();
+
+  bool Changed = false;
+  for (MachineBasicBlock &MBB : MF)
+    Changed |= runOnMachineBasicBlock(MBB);
+
+  return Changed;
+}
+
+/// Returns an instance of the Insert VSETVLI pass.
+FunctionPass *llvm::createRISCVInsertVSETVLIPass() {
+  return new RISCVInsertVSETVLI();
+}
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
@@ -619,8 +619,6 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -637,9 +635,7 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = "$rd = $merge";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -654,8 +650,6 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -672,9 +666,7 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = "$rd = $merge";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -690,8 +682,6 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -710,9 +700,7 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = !if(!eq(EarlyClobber, 1), "@earlyclobber $rd, $rd = $merge", "$rd = $merge");
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -727,8 +715,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -743,8 +729,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -758,8 +742,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -774,8 +756,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -790,8 +770,6 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -804,8 +782,6 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -819,9 +795,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints ="$rd = $merge";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -836,8 +810,6 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   // BaseInstr is not used in RISCVExpandPseudoInsts pass.
@@ -853,9 +825,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = Constraint;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -870,9 +840,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = Join<[Constraint, "$rd = $merge"], ",">.ret;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -887,8 +855,6 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -906,9 +872,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = "@earlyclobber $rd, $rd = $merge";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -925,9 +889,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = Constraint;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -943,8 +905,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -960,8 +920,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -979,9 +937,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = Join<[Constraint, "$rd = $merge"], ",">.ret;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -1001,9 +957,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = Join<[Constraint, "$rd = $merge"], ",">.ret;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -1025,9 +979,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = Constraint;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 0;
@@ -1047,9 +999,7 @@
   let mayLoad = 0;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = Join<[Constraint, "$rd = $rs3"], ",">.ret;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -1068,9 +1018,7 @@
   let mayLoad = 1;
   let mayStore = 1;
   let hasSideEffects = 1;
-  let usesCustomInserter = 1;
   let Constraints = "$vd_wd = $vd";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -1088,9 +1036,7 @@
   let mayLoad = 1;
   let mayStore = 1;
   let hasSideEffects = 1;
-  let usesCustomInserter = 1;
   let Constraints = "$vd_wd = $vd";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -1131,8 +1077,6 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -1148,9 +1092,7 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = "$rd = $merge";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -1166,8 +1108,6 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -1183,9 +1123,7 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   let Constraints = "$rd = $merge";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -1201,11 +1139,9 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   // For vector indexed segment loads, the destination vector register groups
   // cannot overlap the source vector register group
   let Constraints = "@earlyclobber $rd";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -1222,11 +1158,9 @@
   let mayLoad = 1;
   let mayStore = 0;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
   // For vector indexed segment loads, the destination vector register groups
   // cannot overlap the source vector register group
   let Constraints = "@earlyclobber $rd, $rd = $merge";
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasMergeOp = 1;
@@ -1241,8 +1175,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -1258,8 +1190,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -1273,8 +1203,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -1290,8 +1218,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -1307,8 +1233,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let HasDummyMask = 1;
@@ -1325,8 +1249,6 @@
   let mayLoad = 0;
   let mayStore = 1;
   let hasSideEffects = 0;
-  let usesCustomInserter = 1;
-  let Uses = [VL, VTYPE];
   let HasVLOp = 1;
   let HasSEWOp = 1;
   let BaseInstr = !cast<Instruction>(PseudoToVInst<NAME>.VInst);
@@ -3521,7 +3443,7 @@
 //===----------------------------------------------------------------------===//
 // 13.2. Vector Single-Width Averaging Add and Subtract
 //===----------------------------------------------------------------------===//
-let Uses = [VL, VTYPE, VXRM], hasSideEffects = 1 in {
+let Uses = [VXRM], hasSideEffects = 1 in {
   defm PseudoVAADDU      : VPseudoBinaryV_VV_VX;
   defm PseudoVAADD       : VPseudoBinaryV_VV_VX;
   defm PseudoVASUBU      : VPseudoBinaryV_VV_VX;
@@ -3531,14 +3453,14 @@
 //===----------------------------------------------------------------------===//
 // 13.3. Vector Single-Width Fractional Multiply with Rounding and Saturation
 //===----------------------------------------------------------------------===//
-let Uses = [VL, VTYPE, VXRM], Defs = [VXSAT], hasSideEffects = 1 in {
+let Uses = [VXRM], Defs = [VXSAT], hasSideEffects = 1 in {
   defm PseudoVSMUL      : VPseudoBinaryV_VV_VX;
 }
 
 //===----------------------------------------------------------------------===//
 // 13.4. Vector Single-Width Scaling Shift Instructions
 //===----------------------------------------------------------------------===//
-let Uses = [VL, VTYPE, VXRM], hasSideEffects = 1 in {
+let Uses = [VXRM], hasSideEffects = 1 in {
   defm PseudoVSSRL        : VPseudoBinaryV_VV_VX_VI<uimm5>;
   defm PseudoVSSRA        : VPseudoBinaryV_VV_VX_VI<uimm5>;
 }
@@ -3546,7 +3468,7 @@
 //===----------------------------------------------------------------------===//
 // 13.5. Vector Narrowing Fixed-Point Clip Instructions
 //===----------------------------------------------------------------------===//
-let Uses = [VL, VTYPE, VXRM], Defs = [VXSAT], hasSideEffects = 1 in {
+let Uses = [VXRM], Defs = [VXSAT], hasSideEffects = 1 in {
   defm PseudoVNCLIP     : VPseudoBinaryV_WV_WX_WI;
   defm PseudoVNCLIPU    : VPseudoBinaryV_WV_WX_WI;
 }
@@ -3792,8 +3714,7 @@
 //===----------------------------------------------------------------------===//
 
 let Predicates = [HasStdExtV] in {
-let mayLoad = 0, mayStore = 0, hasSideEffects = 0, usesCustomInserter = 1,
-    Uses = [VL, VTYPE] in {
+let mayLoad = 0, mayStore = 0, hasSideEffects = 0 in {
   foreach m = MxList.m in {
     let VLMul = m.value in {
       let HasSEWOp = 1, BaseInstr = VMV_X_S in
@@ -3816,8 +3737,7 @@
 //===----------------------------------------------------------------------===//
 
 let Predicates = [HasStdExtV, HasStdExtF] in {
-let mayLoad = 0, mayStore = 0, hasSideEffects = 0, usesCustomInserter = 1,
-    Uses = [VL, VTYPE] in {
+let mayLoad = 0, mayStore = 0, hasSideEffects = 0 in {
   foreach m = MxList.m in {
     foreach f = FPList.fpinfo in {
       let VLMul = m.value in {
diff --git a/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp b/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp
--- a/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp
@@ -39,7 +39,7 @@
   initializeGlobalISel(*PR);
   initializeRISCVMergeBaseOffsetOptPass(*PR);
   initializeRISCVExpandPseudoPass(*PR);
-  initializeRISCVCleanupVSETVLIPass(*PR);
+  initializeRISCVInsertVSETVLIPass(*PR);
 }
 
 static StringRef computeDataLayout(const Triple &TT) {
@@ -191,8 +191,7 @@
 }
 
 void RISCVPassConfig::addPreRegAlloc() {
-  if (TM->getOptLevel() != CodeGenOpt::None) {
+  if (TM->getOptLevel() != CodeGenOpt::None)
     addPass(createRISCVMergeBaseOffsetOptPass());
-    addPass(createRISCVCleanupVSETVLIPass());
-  }
+  addPass(createRISCVInsertVSETVLIPass());
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/add-vsetvli-gpr.mir b/llvm/test/CodeGen/RISCV/rvv/add-vsetvli-gpr.mir
deleted file mode 100644
--- a/llvm/test/CodeGen/RISCV/rvv/add-vsetvli-gpr.mir
+++ /dev/null
@@ -1,56 +0,0 @@
-# RUN: llc -mtriple riscv64 -mattr=+experimental-v %s  \
-# RUN:     -start-before=finalize-isel -stop-after=finalize-isel -o - \
-# RUN:     | FileCheck --check-prefix=POST-INSERTER %s
-
-# RUN: llc -mtriple riscv64 -mattr=+experimental-v %s  \
-# RUN:     -start-before=finalize-isel -o - \
-# RUN:     | FileCheck --check-prefix=CODEGEN %s
-
---- |
-  define void @vadd_vint64m1(
-            <vscale x 1 x i64> *%pc,
-            <vscale x 1 x i64> *%pa,
-            <vscale x 1 x i64> *%pb,
-            i64 %vl)
-  {
-    ret void
-  }
-...
----
-name: vadd_vint64m1
-tracksRegLiveness: true
-body:             |
-  bb.0 (%ir-block.0):
-    liveins: $x10, $x11, $x12, $x13
-
-    %3:gpr = COPY $x13
-    %2:gpr = COPY $x12
-    %1:gpr = COPY $x11
-    %0:gpr = COPY $x10
-    %4:vr = PseudoVLE64_V_M1 %1, %3, 6, implicit $vl, implicit $vtype :: (load unknown-size from %ir.pa, align 8)
-    %5:vr = PseudoVLE64_V_M1 %2, %3, 6, implicit $vl, implicit $vtype :: (load unknown-size from %ir.pb, align 8)
-    %6:vr = PseudoVADD_VV_M1 killed %4, killed %5, %3, 6, implicit $vl, implicit $vtype
-    PseudoVSE64_V_M1 killed %6, %0, %3, 6, implicit $vl, implicit $vtype :: (store unknown-size into %ir.pc, align 8)
-    PseudoRET
-
-...
-
-# POST-INSERTER: %0:gpr = COPY $x13
-# POST-INSERTER: %1:gpr = COPY $x12
-# POST-INSERTER: %2:gpr = COPY $x11
-# POST-INSERTER: %3:gpr = COPY $x10
-# POST-INSERTER: dead %7:gpr = PseudoVSETVLI %0, 88, implicit-def $vl, implicit-def $vtype
-# POST-INSERTER: %4:vr = PseudoVLE64_V_M1 %2, $noreg, 6, implicit $vl, implicit $vtype :: (load unknown-size from %ir.pa, align 8)
-# POST-INSERTER: dead %8:gpr = PseudoVSETVLI %0, 88, implicit-def $vl, implicit-def $vtype
-# POST-INSERTER: %5:vr = PseudoVLE64_V_M1 %1, $noreg, 6, implicit $vl, implicit $vtype :: (load unknown-size from %ir.pb, align 8)
-# POST-INSERTER: dead %9:gpr = PseudoVSETVLI %0, 88, implicit-def $vl, implicit-def $vtype
-# POST-INSERTER: %6:vr = PseudoVADD_VV_M1 killed %4, killed %5, $noreg, 6, implicit $vl, implicit $vtype
-# POST-INSERTER: dead %10:gpr = PseudoVSETVLI %0, 88, implicit-def $vl, implicit-def $vtype
-# POST-INSERTER: PseudoVSE64_V_M1 killed %6, %3, $noreg, 6, implicit $vl, implicit $vtype :: (store unknown-size into %ir.pc, align 8)
-
-# CODEGEN: vsetvli	a3, a3, e64,m1,ta,mu
-# CODEGEN-NEXT: vle64.v	v25, (a1)
-# CODEGEN-NEXT: vle64.v	v26, (a2)
-# CODEGEN-NEXT: vadd.vv	v25, v25, v26
-# CODEGEN-NEXT: vse64.v	v25, (a0)
-# CODEGEN-NEXT: ret
diff --git a/llvm/test/CodeGen/RISCV/rvv/add-vsetvli-vlmax.ll b/llvm/test/CodeGen/RISCV/rvv/add-vsetvli-vlmax.ll
deleted file mode 100644
--- a/llvm/test/CodeGen/RISCV/rvv/add-vsetvli-vlmax.ll
+++ /dev/null
@@ -1,32 +0,0 @@
-; This test shows the evolution of RVV pseudo instructions within isel.
-
-; RUN: llc -mtriple riscv64 -mattr=+experimental-v %s -o %t.pre.mir \
-; RUN:     -stop-before=finalize-isel
-; RUN: cat %t.pre.mir | FileCheck --check-prefix=PRE-INSERTER %s
-
-; RUN: llc -mtriple riscv64 -mattr=+experimental-v %t.pre.mir -o %t.post.mir \
-; RUN:     -start-before=finalize-isel -stop-after=finalize-isel
-; RUN: cat %t.post.mir | FileCheck --check-prefix=POST-INSERTER %s
-
-define void @vadd_vint64m1(
-          <vscale x 1 x i64> *%pc,
-          <vscale x 1 x i64> *%pa,
-          <vscale x 1 x i64> *%pb)
-{
-  %va = load <vscale x 1 x i64>, <vscale x 1 x i64>* %pa
-  %vb = load <vscale x 1 x i64>, <vscale x 1 x i64>* %pb
-  %vc = add <vscale x 1 x i64> %va, %vb
-  store <vscale x 1 x i64> %vc, <vscale x 1 x i64> *%pc
-  ret void
-}
-
-; PRE-INSERTER: %3:vr = VL1RE64_V %1 :: (load unknown-size from %ir.pa, align 8)
-; PRE-INSERTER: %4:vr = VL1RE64_V %2 :: (load unknown-size from %ir.pb, align 8)
-; PRE-INSERTER: %5:vr = PseudoVADD_VV_M1 killed %3, killed %4, $x0, 6, implicit $vl, implicit $vtype
-; PRE-INSERTER:  VS1R_V killed %5, %0 :: (store unknown-size into %ir.pc, align 8)
-
-; POST-INSERTER: %3:vr = VL1RE64_V %1 :: (load unknown-size from %ir.pa, align 8)
-; POST-INSERTER: %4:vr = VL1RE64_V %2 :: (load unknown-size from %ir.pb, align 8)
-; POST-INSERTER: dead %6:gpr = PseudoVSETVLI $x0, 88, implicit-def $vl, implicit-def $vtype
-; POST-INSERTER: %5:vr = PseudoVADD_VV_M1 killed %3, killed %4, $noreg, 6, implicit $vl, implicit $vtype
-; POST-INSERTER: VS1R_V killed %5, %0 :: (store unknown-size into %ir.pc, align 8)
diff --git a/llvm/test/CodeGen/RISCV/rvv/addi-scalable-offset.mir b/llvm/test/CodeGen/RISCV/rvv/addi-scalable-offset.mir
--- a/llvm/test/CodeGen/RISCV/rvv/addi-scalable-offset.mir
+++ b/llvm/test/CodeGen/RISCV/rvv/addi-scalable-offset.mir
@@ -1,5 +1,5 @@
 # NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -march=riscv64 -stop-after=prologepilog %s -o - 2>&1 | FileCheck %s
+# RUN: llc -march=riscv64 -mattr=+experimental-v -stop-after=prologepilog %s -o - 2>&1 | FileCheck %s
 
 --- |
   define void @add_scalable_offset(
@@ -55,7 +55,7 @@
     ; CHECK: PseudoRET
     %1:gpr = COPY $x11
     %0:gpr = COPY $x10
-    %2:vr = PseudoVLE64_V_M1 %0, %1, 6, implicit $vl, implicit $vtype :: (load unknown-size from %ir.pa, align 8)
+    %2:vr = PseudoVLE64_V_M1 %0, %1, 6 :: (load unknown-size from %ir.pa, align 8)
     %3:gpr = ADDI %stack.2, 0
     VS1R_V killed %2:vr, %3:gpr
     PseudoRET
diff --git a/llvm/test/CodeGen/RISCV/rvv/cleanup-vsetivli.mir b/llvm/test/CodeGen/RISCV/rvv/cleanup-vsetivli.mir
deleted file mode 100644
--- a/llvm/test/CodeGen/RISCV/rvv/cleanup-vsetivli.mir
+++ /dev/null
@@ -1,46 +0,0 @@
-# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc %s -mtriple=riscv64 -run-pass=riscv-cleanup-vsetvli -o - | FileCheck %s
-
-# Make sure we don't combine these VSET{I}VLIs in the cleanup pass. We could not
-# differentiate AVL values if the opcode of the previous one is different from
-# current one.
-
---- |
-  ; ModuleID = '../llvm/test/CodeGen/RISCV/rvv/add-vsetivli.ll'
-  source_filename = "../llvm/test/CodeGen/RISCV/rvv/add-vsetivli.ll"
-  target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n64-S128"
-  target triple = "riscv64"
-
-  define void @cleanup_vsetivli() #0 {
-    ret void
-  }
-
-  attributes #0 = { "target-features"="+experimental-v" }
-
-...
----
-name:            cleanup_vsetivli
-alignment:       4
-tracksRegLiveness: true
-registers:
-  - { id: 0, class: gpr }
-frameInfo:
-  maxAlignment:    1
-machineFunctionInfo: {}
-body:             |
-  bb.0 (%ir-block.0):
-    ; CHECK-LABEL: name: cleanup_vsetivli
-    ; CHECK: dead %0:gpr = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: dead %1:gpr = PseudoVSETIVLI 5, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: dead %3:gpr = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: dead %5:gpr = PseudoVSETIVLI 5, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: PseudoRET
-    dead %0:gpr  = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    dead %1:gpr  = PseudoVSETIVLI 5, 12, implicit-def $vl, implicit-def $vtype
-    dead %2:gpr  = PseudoVSETIVLI 5, 12, implicit-def $vl, implicit-def $vtype
-    dead %3:gpr  = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    dead %4:gpr  = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    dead %5:gpr  = PseudoVSETIVLI 5, 12, implicit-def $vl, implicit-def $vtype
-    PseudoRET
-
-...
diff --git a/llvm/test/CodeGen/RISCV/rvv/cleanup-vsetvli.mir b/llvm/test/CodeGen/RISCV/rvv/cleanup-vsetvli.mir
deleted file mode 100644
--- a/llvm/test/CodeGen/RISCV/rvv/cleanup-vsetvli.mir
+++ /dev/null
@@ -1,79 +0,0 @@
-# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc %s -mtriple=riscv64 -run-pass=riscv-cleanup-vsetvli -o - | FileCheck %s
-
---- |
-  ; ModuleID = '../llvm/test/CodeGen/RISCV/rvv/add-vsetvli-vlmax.ll'
-  source_filename = "../llvm/test/CodeGen/RISCV/rvv/add-vsetvli-vlmax.ll"
-  target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n64-S128"
-  target triple = "riscv64"
-
-  define void @cleanup_vsetvli0() #0 {
-    ret void
-  }
-
-  define void @cleanup_vsetvli1() #0 {
-    ret void
-  }
-
-  attributes #0 = { "target-features"="+experimental-v" }
-
-...
----
-# Make sure we don't combine these two VSETVLIs in the cleanup pass. The first
-# keeps the previous value of VL, the second sets it to VLMAX. We can't remove
-# the first since we can't tell if this is a change of VL.
-name:            cleanup_vsetvli0
-alignment:       4
-tracksRegLiveness: true
-registers:
-  - { id: 0, class: gpr }
-frameInfo:
-  maxAlignment:    1
-machineFunctionInfo: {}
-body:             |
-  bb.0 (%ir-block.0):
-    ; CHECK-LABEL: name: cleanup_vsetvli0
-    ; CHECK: dead $x0 = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: dead %0:gpr = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: PseudoRET
-    dead $x0  = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    dead %0:gpr  = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    PseudoRET
-
-...
----
-# 1. Ensure we can remove the second VSETVLI which takes its AVL from the first VSETVLI.
-# 2. Ensure we can remove the fourth VSETVLI which takes its AVL from the VSETIVLI.
-# 3. Make sure we don't combine the latter two VSETVLIs; the first outputs to a
-# physical register which is clobbered by a later instruction.
-name:            cleanup_vsetvli1
-alignment:       4
-tracksRegLiveness: true
-registers:
-  - { id: 0, class: gpr }
-frameInfo:
-  maxAlignment:    1
-machineFunctionInfo: {}
-body:             |
-  bb.0 (%ir-block.0):
-    liveins: $x3
-    ; CHECK-LABEL: name: cleanup_vsetvli1
-    ; CHECK: liveins: $x3
-    ; CHECK: [[PseudoVSETVLI:%[0-9]+]]:gpr = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: [[PseudoVSETIVLI:%[0-9]+]]:gpr = PseudoVSETIVLI 4, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: $x1 = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: $x1 = COPY $x3
-    ; CHECK: dead %4:gpr = PseudoVSETVLI $x1, 12, implicit-def $vl, implicit-def $vtype
-    ; CHECK: PseudoRET
-    %0:gpr  = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    dead %1:gpr  = PseudoVSETVLI %0, 12, implicit-def $vl, implicit-def $vtype
-
-    %2:gpr  = PseudoVSETIVLI 4, 12, implicit-def $vl, implicit-def $vtype
-    dead %3:gpr  = PseudoVSETVLI %2, 12, implicit-def $vl, implicit-def $vtype
-
-    $x1  = PseudoVSETVLI $x0, 12, implicit-def $vl, implicit-def $vtype
-    $x1 = COPY $x3
-    dead %4:gpr  = PseudoVSETVLI $x1, 12, implicit-def $vl, implicit-def $vtype
-    PseudoRET
-
-...
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll
@@ -3667,11 +3667,12 @@
 ; LMULMAX2-RV32-NEXT:    addi a3, a1, 819
 ; LMULMAX2-RV32-NEXT:    lui a1, 61681
 ; LMULMAX2-RV32-NEXT:    addi a7, a1, -241
-; LMULMAX2-RV32-NEXT:    lui a1, 4112
-; LMULMAX2-RV32-NEXT:    addi a2, a1, 257
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT:    lui a2, 4112
+; LMULMAX2-RV32-NEXT:    addi a2, a2, 257
 ; LMULMAX2-RV32-NEXT:    bnez a5, .LBB3_2
 ; LMULMAX2-RV32-NEXT:  # %bb.1:
+; LMULMAX2-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v25
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
@@ -3726,12 +3727,13 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v25, v25, 1
 ; LMULMAX2-RV32-NEXT:    vsrl.vx v26, v25, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v25
-; LMULMAX2-RV32-NEXT:    bnez a1, .LBB3_5
+; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v26
+; LMULMAX2-RV32-NEXT:    bnez a5, .LBB3_5
 ; LMULMAX2-RV32-NEXT:  # %bb.4:
-; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
-; LMULMAX2-RV32-NEXT:    or a1, a5, a1
+; LMULMAX2-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
+; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -3756,8 +3758,8 @@
 ; LMULMAX2-RV32-NEXT:    addi a1, a1, 32
 ; LMULMAX2-RV32-NEXT:    j .LBB3_6
 ; LMULMAX2-RV32-NEXT:  .LBB3_5:
-; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
-; LMULMAX2-RV32-NEXT:    or a1, a1, a5
+; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
+; LMULMAX2-RV32-NEXT:    or a1, a5, a1
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -3900,11 +3902,12 @@
 ; LMULMAX1-RV32-NEXT:    addi a3, a1, 819
 ; LMULMAX1-RV32-NEXT:    lui a1, 61681
 ; LMULMAX1-RV32-NEXT:    addi a7, a1, -241
-; LMULMAX1-RV32-NEXT:    lui a1, 4112
-; LMULMAX1-RV32-NEXT:    addi a2, a1, 257
-; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT:    lui a2, 4112
+; LMULMAX1-RV32-NEXT:    addi a2, a2, 257
 ; LMULMAX1-RV32-NEXT:    bnez a5, .LBB3_2
 ; LMULMAX1-RV32-NEXT:  # %bb.1:
+; LMULMAX1-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX1-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 2
@@ -3959,12 +3962,13 @@
 ; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vslidedown.vi v25, v25, 1
 ; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v25, a6
-; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
-; LMULMAX1-RV32-NEXT:    vmv.x.s a5, v25
-; LMULMAX1-RV32-NEXT:    bnez a1, .LBB3_5
+; LMULMAX1-RV32-NEXT:    vmv.x.s a5, v26
+; LMULMAX1-RV32-NEXT:    bnez a5, .LBB3_5
 ; LMULMAX1-RV32-NEXT:  # %bb.4:
-; LMULMAX1-RV32-NEXT:    srli a1, a5, 1
-; LMULMAX1-RV32-NEXT:    or a1, a5, a1
+; LMULMAX1-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT:    srli a5, a1, 1
+; LMULMAX1-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX1-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 4
@@ -3989,8 +3993,8 @@
 ; LMULMAX1-RV32-NEXT:    addi a1, a1, 32
 ; LMULMAX1-RV32-NEXT:    j .LBB3_6
 ; LMULMAX1-RV32-NEXT:  .LBB3_5:
-; LMULMAX1-RV32-NEXT:    srli a5, a1, 1
-; LMULMAX1-RV32-NEXT:    or a1, a1, a5
+; LMULMAX1-RV32-NEXT:    srli a1, a5, 1
+; LMULMAX1-RV32-NEXT:    or a1, a5, a1
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX1-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 4
@@ -11120,11 +11124,12 @@
 ; LMULMAX2-RV32-NEXT:    addi a3, a1, 819
 ; LMULMAX2-RV32-NEXT:    lui a1, 61681
 ; LMULMAX2-RV32-NEXT:    addi a7, a1, -241
-; LMULMAX2-RV32-NEXT:    lui a1, 4112
-; LMULMAX2-RV32-NEXT:    addi a2, a1, 257
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT:    lui a2, 4112
+; LMULMAX2-RV32-NEXT:    addi a2, a2, 257
 ; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_2
 ; LMULMAX2-RV32-NEXT:  # %bb.1:
+; LMULMAX2-RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
@@ -11179,12 +11184,13 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v28, v26, 3
 ; LMULMAX2-RV32-NEXT:    vsrl.vx v30, v28, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v30
-; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v28
-; LMULMAX2-RV32-NEXT:    bnez a1, .LBB7_5
+; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v30
+; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_5
 ; LMULMAX2-RV32-NEXT:  # %bb.4:
-; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
-; LMULMAX2-RV32-NEXT:    or a1, a5, a1
+; LMULMAX2-RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
+; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
+; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -11209,8 +11215,8 @@
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, 32
 ; LMULMAX2-RV32-NEXT:    j .LBB7_6
 ; LMULMAX2-RV32-NEXT:  .LBB7_5:
-; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
-; LMULMAX2-RV32-NEXT:    or a1, a1, a5
+; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
+; LMULMAX2-RV32-NEXT:    or a1, a5, a1
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -11237,12 +11243,13 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v28, v26, 2
 ; LMULMAX2-RV32-NEXT:    vsrl.vx v30, v28, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v30
-; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v28
-; LMULMAX2-RV32-NEXT:    bnez a1, .LBB7_8
+; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v30
+; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_8
 ; LMULMAX2-RV32-NEXT:  # %bb.7:
-; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
-; LMULMAX2-RV32-NEXT:    or a1, a5, a1
+; LMULMAX2-RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
+; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
+; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -11267,8 +11274,8 @@
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, 32
 ; LMULMAX2-RV32-NEXT:    j .LBB7_9
 ; LMULMAX2-RV32-NEXT:  .LBB7_8:
-; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
-; LMULMAX2-RV32-NEXT:    or a1, a1, a5
+; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
+; LMULMAX2-RV32-NEXT:    or a1, a5, a1
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -11295,12 +11302,13 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v26, v26, 1
 ; LMULMAX2-RV32-NEXT:    vsrl.vx v28, v26, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v26
-; LMULMAX2-RV32-NEXT:    bnez a1, .LBB7_11
+; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v28
+; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_11
 ; LMULMAX2-RV32-NEXT:  # %bb.10:
-; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
-; LMULMAX2-RV32-NEXT:    or a1, a5, a1
+; LMULMAX2-RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
+; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -11325,8 +11333,8 @@
 ; LMULMAX2-RV32-NEXT:    addi a1, a1, 32
 ; LMULMAX2-RV32-NEXT:    j .LBB7_12
 ; LMULMAX2-RV32-NEXT:  .LBB7_11:
-; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
-; LMULMAX2-RV32-NEXT:    or a1, a1, a5
+; LMULMAX2-RV32-NEXT:    srli a1, a5, 1
+; LMULMAX2-RV32-NEXT:    or a1, a5, a1
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 2
 ; LMULMAX2-RV32-NEXT:    or a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 4
@@ -11544,13 +11552,14 @@
 ; LMULMAX1-RV32-NEXT:    addi a4, a2, 819
 ; LMULMAX1-RV32-NEXT:    lui a2, 61681
 ; LMULMAX1-RV32-NEXT:    addi t0, a2, -241
-; LMULMAX1-RV32-NEXT:    lui a2, 4112
-; LMULMAX1-RV32-NEXT:    addi a3, a2, 257
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT:    lui a3, 4112
+; LMULMAX1-RV32-NEXT:    addi a3, a3, 257
 ; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_2
 ; LMULMAX1-RV32-NEXT:  # %bb.1:
-; LMULMAX1-RV32-NEXT:    srli a1, a2, 1
-; LMULMAX1-RV32-NEXT:    or a1, a2, a1
+; LMULMAX1-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
+; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 2
 ; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 4
@@ -11604,11 +11613,12 @@
 ; LMULMAX1-RV32-NEXT:    vslidedown.vi v26, v26, 1
 ; LMULMAX1-RV32-NEXT:    vsrl.vx v27, v26, a7
 ; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v27
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v26
 ; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_5
 ; LMULMAX1-RV32-NEXT:  # %bb.4:
-; LMULMAX1-RV32-NEXT:    srli a1, a2, 1
-; LMULMAX1-RV32-NEXT:    or a1, a2, a1
+; LMULMAX1-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
+; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 2
 ; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 4
@@ -11663,11 +11673,12 @@
 ; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v25, a7
 ; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v25
 ; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_8
 ; LMULMAX1-RV32-NEXT:  # %bb.7:
-; LMULMAX1-RV32-NEXT:    srli a1, a2, 1
-; LMULMAX1-RV32-NEXT:    or a1, a2, a1
+; LMULMAX1-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
+; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 2
 ; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 4
@@ -11721,11 +11732,12 @@
 ; LMULMAX1-RV32-NEXT:    vslidedown.vi v25, v25, 1
 ; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v25, a7
 ; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v25
 ; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_11
 ; LMULMAX1-RV32-NEXT:  # %bb.10:
-; LMULMAX1-RV32-NEXT:    srli a1, a2, 1
-; LMULMAX1-RV32-NEXT:    or a1, a2, a1
+; LMULMAX1-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
+; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 2
 ; LMULMAX1-RV32-NEXT:    or a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 4
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll
@@ -2538,9 +2538,6 @@
 ; LMULMAX2-RV32-NEXT:    sw zero, 12(sp)
 ; LMULMAX2-RV32-NEXT:    sw zero, 4(sp)
 ; LMULMAX2-RV32-NEXT:    addi a6, zero, 32
-; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
-; LMULMAX2-RV32-NEXT:    vsrl.vx v26, v25, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v26
 ; LMULMAX2-RV32-NEXT:    lui a1, 349525
 ; LMULMAX2-RV32-NEXT:    addi a4, a1, 1365
 ; LMULMAX2-RV32-NEXT:    lui a1, 209715
@@ -2548,13 +2545,16 @@
 ; LMULMAX2-RV32-NEXT:    lui a1, 61681
 ; LMULMAX2-RV32-NEXT:    addi a7, a1, -241
 ; LMULMAX2-RV32-NEXT:    lui a2, 4112
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v25
 ; LMULMAX2-RV32-NEXT:    addi a2, a2, 257
-; LMULMAX2-RV32-NEXT:    bnez a1, .LBB3_2
+; LMULMAX2-RV32-NEXT:    bnez a5, .LBB3_2
 ; LMULMAX2-RV32-NEXT:  # %bb.1:
-; LMULMAX2-RV32-NEXT:    addi a1, a5, -1
-; LMULMAX2-RV32-NEXT:    not a5, a5
-; LMULMAX2-RV32-NEXT:    and a1, a5, a1
+; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT:    vsrl.vx v26, v25, a6
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
+; LMULMAX2-RV32-NEXT:    not a1, a1
+; LMULMAX2-RV32-NEXT:    and a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX2-RV32-NEXT:    and a5, a5, a4
 ; LMULMAX2-RV32-NEXT:    sub a1, a1, a5
@@ -2570,9 +2570,9 @@
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, 32
 ; LMULMAX2-RV32-NEXT:    j .LBB3_3
 ; LMULMAX2-RV32-NEXT:  .LBB3_2:
-; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
-; LMULMAX2-RV32-NEXT:    not a1, a1
-; LMULMAX2-RV32-NEXT:    and a1, a1, a5
+; LMULMAX2-RV32-NEXT:    addi a1, a5, -1
+; LMULMAX2-RV32-NEXT:    not a5, a5
+; LMULMAX2-RV32-NEXT:    and a1, a5, a1
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX2-RV32-NEXT:    and a5, a5, a4
 ; LMULMAX2-RV32-NEXT:    sub a1, a1, a5
@@ -2590,10 +2590,11 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v25, v25, 1
 ; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v25
-; LMULMAX2-RV32-NEXT:    vsrl.vx v25, v25, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v25
 ; LMULMAX2-RV32-NEXT:    bnez a5, .LBB3_5
 ; LMULMAX2-RV32-NEXT:  # %bb.4:
+; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT:    vsrl.vx v25, v25, a6
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v25
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
 ; LMULMAX2-RV32-NEXT:    not a1, a1
 ; LMULMAX2-RV32-NEXT:    and a1, a1, a5
@@ -2719,9 +2720,6 @@
 ; LMULMAX1-RV32-NEXT:    sw zero, 12(sp)
 ; LMULMAX1-RV32-NEXT:    sw zero, 4(sp)
 ; LMULMAX1-RV32-NEXT:    addi a6, zero, 32
-; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
-; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v25, a6
-; LMULMAX1-RV32-NEXT:    vmv.x.s a5, v26
 ; LMULMAX1-RV32-NEXT:    lui a1, 349525
 ; LMULMAX1-RV32-NEXT:    addi a4, a1, 1365
 ; LMULMAX1-RV32-NEXT:    lui a1, 209715
@@ -2729,13 +2727,16 @@
 ; LMULMAX1-RV32-NEXT:    lui a1, 61681
 ; LMULMAX1-RV32-NEXT:    addi a7, a1, -241
 ; LMULMAX1-RV32-NEXT:    lui a2, 4112
-; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT:    vmv.x.s a5, v25
 ; LMULMAX1-RV32-NEXT:    addi a2, a2, 257
-; LMULMAX1-RV32-NEXT:    bnez a1, .LBB3_2
+; LMULMAX1-RV32-NEXT:    bnez a5, .LBB3_2
 ; LMULMAX1-RV32-NEXT:  # %bb.1:
-; LMULMAX1-RV32-NEXT:    addi a1, a5, -1
-; LMULMAX1-RV32-NEXT:    not a5, a5
-; LMULMAX1-RV32-NEXT:    and a1, a5, a1
+; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v25, a6
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT:    addi a5, a1, -1
+; LMULMAX1-RV32-NEXT:    not a1, a1
+; LMULMAX1-RV32-NEXT:    and a1, a1, a5
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX1-RV32-NEXT:    and a5, a5, a4
 ; LMULMAX1-RV32-NEXT:    sub a1, a1, a5
@@ -2751,9 +2752,9 @@
 ; LMULMAX1-RV32-NEXT:    addi a5, a1, 32
 ; LMULMAX1-RV32-NEXT:    j .LBB3_3
 ; LMULMAX1-RV32-NEXT:  .LBB3_2:
-; LMULMAX1-RV32-NEXT:    addi a5, a1, -1
-; LMULMAX1-RV32-NEXT:    not a1, a1
-; LMULMAX1-RV32-NEXT:    and a1, a1, a5
+; LMULMAX1-RV32-NEXT:    addi a1, a5, -1
+; LMULMAX1-RV32-NEXT:    not a5, a5
+; LMULMAX1-RV32-NEXT:    and a1, a5, a1
 ; LMULMAX1-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX1-RV32-NEXT:    and a5, a5, a4
 ; LMULMAX1-RV32-NEXT:    sub a1, a1, a5
@@ -2771,10 +2772,11 @@
 ; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vslidedown.vi v25, v25, 1
 ; LMULMAX1-RV32-NEXT:    vmv.x.s a5, v25
-; LMULMAX1-RV32-NEXT:    vsrl.vx v25, v25, a6
-; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
 ; LMULMAX1-RV32-NEXT:    bnez a5, .LBB3_5
 ; LMULMAX1-RV32-NEXT:  # %bb.4:
+; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vsrl.vx v25, v25, a6
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
 ; LMULMAX1-RV32-NEXT:    addi a5, a1, -1
 ; LMULMAX1-RV32-NEXT:    not a1, a1
 ; LMULMAX1-RV32-NEXT:    and a1, a1, a5
@@ -7647,9 +7649,6 @@
 ; LMULMAX2-RV32-NEXT:    sw zero, 12(sp)
 ; LMULMAX2-RV32-NEXT:    sw zero, 4(sp)
 ; LMULMAX2-RV32-NEXT:    addi a6, zero, 32
-; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
-; LMULMAX2-RV32-NEXT:    vsrl.vx v28, v26, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v28
 ; LMULMAX2-RV32-NEXT:    lui a1, 349525
 ; LMULMAX2-RV32-NEXT:    addi a4, a1, 1365
 ; LMULMAX2-RV32-NEXT:    lui a1, 209715
@@ -7657,13 +7656,16 @@
 ; LMULMAX2-RV32-NEXT:    lui a1, 61681
 ; LMULMAX2-RV32-NEXT:    addi a7, a1, -241
 ; LMULMAX2-RV32-NEXT:    lui a2, 4112
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v26
 ; LMULMAX2-RV32-NEXT:    addi a2, a2, 257
-; LMULMAX2-RV32-NEXT:    bnez a1, .LBB7_2
+; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_2
 ; LMULMAX2-RV32-NEXT:  # %bb.1:
-; LMULMAX2-RV32-NEXT:    addi a1, a5, -1
-; LMULMAX2-RV32-NEXT:    not a5, a5
-; LMULMAX2-RV32-NEXT:    and a1, a5, a1
+; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vsrl.vx v28, v26, a6
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
+; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
+; LMULMAX2-RV32-NEXT:    not a1, a1
+; LMULMAX2-RV32-NEXT:    and a1, a1, a5
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX2-RV32-NEXT:    and a5, a5, a4
 ; LMULMAX2-RV32-NEXT:    sub a1, a1, a5
@@ -7679,9 +7681,9 @@
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, 32
 ; LMULMAX2-RV32-NEXT:    j .LBB7_3
 ; LMULMAX2-RV32-NEXT:  .LBB7_2:
-; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
-; LMULMAX2-RV32-NEXT:    not a1, a1
-; LMULMAX2-RV32-NEXT:    and a1, a1, a5
+; LMULMAX2-RV32-NEXT:    addi a1, a5, -1
+; LMULMAX2-RV32-NEXT:    not a5, a5
+; LMULMAX2-RV32-NEXT:    and a1, a5, a1
 ; LMULMAX2-RV32-NEXT:    srli a5, a1, 1
 ; LMULMAX2-RV32-NEXT:    and a5, a5, a4
 ; LMULMAX2-RV32-NEXT:    sub a1, a1, a5
@@ -7699,10 +7701,11 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v28, v26, 3
 ; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v28
-; LMULMAX2-RV32-NEXT:    vsrl.vx v28, v28, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
 ; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_5
 ; LMULMAX2-RV32-NEXT:  # %bb.4:
+; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vsrl.vx v28, v28, a6
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
 ; LMULMAX2-RV32-NEXT:    not a1, a1
 ; LMULMAX2-RV32-NEXT:    and a1, a1, a5
@@ -7741,10 +7744,11 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v28, v26, 2
 ; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v28
-; LMULMAX2-RV32-NEXT:    vsrl.vx v28, v28, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
 ; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_8
 ; LMULMAX2-RV32-NEXT:  # %bb.7:
+; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vsrl.vx v28, v28, a6
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v28
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
 ; LMULMAX2-RV32-NEXT:    not a1, a1
 ; LMULMAX2-RV32-NEXT:    and a1, a1, a5
@@ -7783,10 +7787,11 @@
 ; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; LMULMAX2-RV32-NEXT:    vslidedown.vi v26, v26, 1
 ; LMULMAX2-RV32-NEXT:    vmv.x.s a5, v26
-; LMULMAX2-RV32-NEXT:    vsrl.vx v26, v26, a6
-; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
 ; LMULMAX2-RV32-NEXT:    bnez a5, .LBB7_11
 ; LMULMAX2-RV32-NEXT:  # %bb.10:
+; LMULMAX2-RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT:    vsrl.vx v26, v26, a6
+; LMULMAX2-RV32-NEXT:    vmv.x.s a1, v26
 ; LMULMAX2-RV32-NEXT:    addi a5, a1, -1
 ; LMULMAX2-RV32-NEXT:    not a1, a1
 ; LMULMAX2-RV32-NEXT:    and a1, a1, a5
@@ -7962,25 +7967,25 @@
 ; LMULMAX1-RV32-NEXT:    .cfi_def_cfa_offset 32
 ; LMULMAX1-RV32-NEXT:    vsetivli a1, 2, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vle64.v v25, (a0)
-; LMULMAX1-RV32-NEXT:    addi a6, a0, 16
-; LMULMAX1-RV32-NEXT:    vle64.v v26, (a6)
+; LMULMAX1-RV32-NEXT:    addi a7, a0, 16
+; LMULMAX1-RV32-NEXT:    vle64.v v26, (a7)
 ; LMULMAX1-RV32-NEXT:    sw zero, 28(sp)
 ; LMULMAX1-RV32-NEXT:    sw zero, 20(sp)
-; LMULMAX1-RV32-NEXT:    addi a7, zero, 32
-; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
-; LMULMAX1-RV32-NEXT:    vsrl.vx v27, v26, a7
-; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v27
-; LMULMAX1-RV32-NEXT:    lui a2, 349525
-; LMULMAX1-RV32-NEXT:    addi a5, a2, 1365
-; LMULMAX1-RV32-NEXT:    lui a2, 209715
-; LMULMAX1-RV32-NEXT:    addi a4, a2, 819
-; LMULMAX1-RV32-NEXT:    lui a2, 61681
-; LMULMAX1-RV32-NEXT:    addi t0, a2, -241
+; LMULMAX1-RV32-NEXT:    addi a6, zero, 32
+; LMULMAX1-RV32-NEXT:    lui a1, 349525
+; LMULMAX1-RV32-NEXT:    addi a5, a1, 1365
+; LMULMAX1-RV32-NEXT:    lui a1, 209715
+; LMULMAX1-RV32-NEXT:    addi a4, a1, 819
+; LMULMAX1-RV32-NEXT:    lui a1, 61681
+; LMULMAX1-RV32-NEXT:    addi t0, a1, -241
 ; LMULMAX1-RV32-NEXT:    lui a3, 4112
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
 ; LMULMAX1-RV32-NEXT:    addi a3, a3, 257
-; LMULMAX1-RV32-NEXT:    bnez a2, .LBB7_2
+; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_2
 ; LMULMAX1-RV32-NEXT:  # %bb.1:
+; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vsrl.vx v27, v26, a6
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v27
 ; LMULMAX1-RV32-NEXT:    addi a2, a1, -1
 ; LMULMAX1-RV32-NEXT:    not a1, a1
 ; LMULMAX1-RV32-NEXT:    and a1, a1, a2
@@ -7999,9 +8004,9 @@
 ; LMULMAX1-RV32-NEXT:    addi a1, a1, 32
 ; LMULMAX1-RV32-NEXT:    j .LBB7_3
 ; LMULMAX1-RV32-NEXT:  .LBB7_2:
-; LMULMAX1-RV32-NEXT:    addi a1, a2, -1
-; LMULMAX1-RV32-NEXT:    not a2, a2
-; LMULMAX1-RV32-NEXT:    and a1, a2, a1
+; LMULMAX1-RV32-NEXT:    addi a2, a1, -1
+; LMULMAX1-RV32-NEXT:    not a1, a1
+; LMULMAX1-RV32-NEXT:    and a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
 ; LMULMAX1-RV32-NEXT:    and a2, a2, a5
 ; LMULMAX1-RV32-NEXT:    sub a1, a1, a2
@@ -8019,13 +8024,14 @@
 ; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vslidedown.vi v26, v26, 1
 ; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
-; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v26, a7
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v26
 ; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_5
 ; LMULMAX1-RV32-NEXT:  # %bb.4:
-; LMULMAX1-RV32-NEXT:    addi a1, a2, -1
-; LMULMAX1-RV32-NEXT:    not a2, a2
-; LMULMAX1-RV32-NEXT:    and a1, a2, a1
+; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v26, a6
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT:    addi a2, a1, -1
+; LMULMAX1-RV32-NEXT:    not a1, a1
+; LMULMAX1-RV32-NEXT:    and a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
 ; LMULMAX1-RV32-NEXT:    and a2, a2, a5
 ; LMULMAX1-RV32-NEXT:    sub a1, a1, a2
@@ -8060,15 +8066,16 @@
 ; LMULMAX1-RV32-NEXT:    sw a1, 24(sp)
 ; LMULMAX1-RV32-NEXT:    sw zero, 12(sp)
 ; LMULMAX1-RV32-NEXT:    sw zero, 4(sp)
-; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
-; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v25, a7
+; LMULMAX1-RV32-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v26
 ; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_8
 ; LMULMAX1-RV32-NEXT:  # %bb.7:
-; LMULMAX1-RV32-NEXT:    addi a1, a2, -1
-; LMULMAX1-RV32-NEXT:    not a2, a2
-; LMULMAX1-RV32-NEXT:    and a1, a2, a1
+; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vsrl.vx v26, v25, a6
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT:    addi a2, a1, -1
+; LMULMAX1-RV32-NEXT:    not a1, a1
+; LMULMAX1-RV32-NEXT:    and a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
 ; LMULMAX1-RV32-NEXT:    and a2, a2, a5
 ; LMULMAX1-RV32-NEXT:    sub a1, a1, a2
@@ -8104,13 +8111,14 @@
 ; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vslidedown.vi v25, v25, 1
 ; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
-; LMULMAX1-RV32-NEXT:    vsrl.vx v25, v25, a7
-; LMULMAX1-RV32-NEXT:    vmv.x.s a2, v25
 ; LMULMAX1-RV32-NEXT:    bnez a1, .LBB7_11
 ; LMULMAX1-RV32-NEXT:  # %bb.10:
-; LMULMAX1-RV32-NEXT:    addi a1, a2, -1
-; LMULMAX1-RV32-NEXT:    not a2, a2
-; LMULMAX1-RV32-NEXT:    and a1, a2, a1
+; LMULMAX1-RV32-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT:    vsrl.vx v25, v25, a6
+; LMULMAX1-RV32-NEXT:    vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT:    addi a2, a1, -1
+; LMULMAX1-RV32-NEXT:    not a1, a1
+; LMULMAX1-RV32-NEXT:    and a1, a1, a2
 ; LMULMAX1-RV32-NEXT:    srli a2, a1, 1
 ; LMULMAX1-RV32-NEXT:    and a2, a2, a5
 ; LMULMAX1-RV32-NEXT:    sub a1, a1, a2
@@ -8149,7 +8157,7 @@
 ; LMULMAX1-RV32-NEXT:    vle32.v v26, (a1)
 ; LMULMAX1-RV32-NEXT:    vsetivli a1, 2, e64,m1,ta,mu
 ; LMULMAX1-RV32-NEXT:    vse64.v v25, (a0)
-; LMULMAX1-RV32-NEXT:    vse64.v v26, (a6)
+; LMULMAX1-RV32-NEXT:    vse64.v v26, (a7)
 ; LMULMAX1-RV32-NEXT:    addi sp, sp, 32
 ; LMULMAX1-RV32-NEXT:    ret
 ;
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-select-fp.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-select-fp.ll
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-select-fp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-select-fp.ll
@@ -7,20 +7,22 @@
 define <2 x half> @select_v2f16(i1 zeroext %c, <2 x half> %a, <2 x half> %b) {
 ; CHECK-LABEL: select_v2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
-; CHECK-NEXT:    vfmv.f.s ft3, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft2, v25
 ; CHECK-NEXT:    bnez a0, .LBB0_2
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.h ft0, ft1
-; CHECK-NEXT:    fmv.h ft2, ft3
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB0_3
 ; CHECK-NEXT:  .LBB0_2:
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 1
+; CHECK-NEXT:  .LBB0_3:
+; CHECK-NEXT:    vfmv.f.s ft1, v25
 ; CHECK-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
-; CHECK-NEXT:    vfmv.v.f v8, ft2
+; CHECK-NEXT:    vfmv.v.f v8, ft1
 ; CHECK-NEXT:    vfmv.s.f v8, ft0
 ; CHECK-NEXT:    ret
   %v = select i1 %c, <2 x half> %a, <2 x half> %b
@@ -31,23 +33,27 @@
 ; CHECK-LABEL: selectcc_v2f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    feq.h a0, fa0, fa1
+; CHECK-NEXT:    bnez a0, .LBB1_2
+; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
+; CHECK-NEXT:    j .LBB1_3
+; CHECK-NEXT:  .LBB1_2:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
+; CHECK-NEXT:  .LBB1_3:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    bnez a0, .LBB1_2
-; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.h ft0, ft1
-; CHECK-NEXT:  .LBB1_2:
 ; CHECK-NEXT:    vsetivli a1, 2, e16,m1,ta,mu
 ; CHECK-NEXT:    vfmv.v.f v25, ft0
-; CHECK-NEXT:    vfmv.f.s ft1, v9
+; CHECK-NEXT:    bnez a0, .LBB1_5
+; CHECK-NEXT:  # %bb.4:
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    j .LBB1_6
+; CHECK-NEXT:  .LBB1_5:
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB1_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.h ft0, ft1
-; CHECK-NEXT:  .LBB1_4:
+; CHECK-NEXT:  .LBB1_6:
 ; CHECK-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
 ; CHECK-NEXT:    vfmv.s.f v25, ft0
 ; CHECK-NEXT:    vmv1r.v v8, v25
@@ -62,44 +68,49 @@
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
-; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB2_2
+; CHECK-NEXT:    bnez a0, .LBB2_3
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.h ft0, ft1
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    fsh ft0, 8(sp)
+; CHECK-NEXT:    beqz a0, .LBB2_4
 ; CHECK-NEXT:  .LBB2_2:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 3
+; CHECK-NEXT:    j .LBB2_5
+; CHECK-NEXT:  .LBB2_3:
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fsh ft0, 8(sp)
+; CHECK-NEXT:    bnez a0, .LBB2_2
+; CHECK-NEXT:  .LBB2_4:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 3
+; CHECK-NEXT:  .LBB2_5:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 3
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB2_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB2_4:
-; CHECK-NEXT:    fsh ft1, 14(sp)
+; CHECK-NEXT:    fsh ft0, 14(sp)
+; CHECK-NEXT:    bnez a0, .LBB2_7
+; CHECK-NEXT:  # %bb.6:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 2
-; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 2
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB2_6
-; CHECK-NEXT:  # %bb.5:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB2_6:
-; CHECK-NEXT:    fsh ft1, 12(sp)
+; CHECK-NEXT:    j .LBB2_8
+; CHECK-NEXT:  .LBB2_7:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    vslidedown.vi v25, v8, 2
+; CHECK-NEXT:  .LBB2_8:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 12(sp)
+; CHECK-NEXT:    bnez a0, .LBB2_10
+; CHECK-NEXT:  # %bb.9:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB2_11
+; CHECK-NEXT:  .LBB2_10:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB2_8
-; CHECK-NEXT:  # %bb.7:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB2_8:
-; CHECK-NEXT:    fsh ft1, 10(sp)
+; CHECK-NEXT:  .LBB2_11:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 10(sp)
 ; CHECK-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vle16.v v8, (a0)
@@ -115,44 +126,49 @@
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    feq.h a0, fa0, fa1
-; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB3_2
+; CHECK-NEXT:    bnez a0, .LBB3_3
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.h ft0, ft1
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    fsh ft0, 8(sp)
+; CHECK-NEXT:    beqz a0, .LBB3_4
 ; CHECK-NEXT:  .LBB3_2:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 3
+; CHECK-NEXT:    j .LBB3_5
+; CHECK-NEXT:  .LBB3_3:
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fsh ft0, 8(sp)
+; CHECK-NEXT:    bnez a0, .LBB3_2
+; CHECK-NEXT:  .LBB3_4:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 3
+; CHECK-NEXT:  .LBB3_5:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 3
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB3_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB3_4:
-; CHECK-NEXT:    fsh ft1, 14(sp)
+; CHECK-NEXT:    fsh ft0, 14(sp)
+; CHECK-NEXT:    bnez a0, .LBB3_7
+; CHECK-NEXT:  # %bb.6:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 2
-; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 2
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB3_6
-; CHECK-NEXT:  # %bb.5:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB3_6:
-; CHECK-NEXT:    fsh ft1, 12(sp)
+; CHECK-NEXT:    j .LBB3_8
+; CHECK-NEXT:  .LBB3_7:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    vslidedown.vi v25, v8, 2
+; CHECK-NEXT:  .LBB3_8:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 12(sp)
+; CHECK-NEXT:    bnez a0, .LBB3_10
+; CHECK-NEXT:  # %bb.9:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB3_11
+; CHECK-NEXT:  .LBB3_10:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB3_8
-; CHECK-NEXT:  # %bb.7:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB3_8:
-; CHECK-NEXT:    fsh ft1, 10(sp)
+; CHECK-NEXT:  .LBB3_11:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 10(sp)
 ; CHECK-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vle16.v v8, (a0)
@@ -168,84 +184,93 @@
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
-; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB4_2
+; CHECK-NEXT:    bnez a0, .LBB4_3
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.h ft0, ft1
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    fsh ft0, 0(sp)
+; CHECK-NEXT:    beqz a0, .LBB4_4
 ; CHECK-NEXT:  .LBB4_2:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 7
+; CHECK-NEXT:    j .LBB4_5
+; CHECK-NEXT:  .LBB4_3:
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fsh ft0, 0(sp)
+; CHECK-NEXT:    bnez a0, .LBB4_2
+; CHECK-NEXT:  .LBB4_4:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 7
+; CHECK-NEXT:  .LBB4_5:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 7
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB4_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB4_4:
-; CHECK-NEXT:    fsh ft1, 14(sp)
+; CHECK-NEXT:    fsh ft0, 14(sp)
+; CHECK-NEXT:    bnez a0, .LBB4_7
+; CHECK-NEXT:  # %bb.6:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 6
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB4_8
+; CHECK-NEXT:  .LBB4_7:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 6
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB4_6
-; CHECK-NEXT:  # %bb.5:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB4_6:
-; CHECK-NEXT:    fsh ft1, 12(sp)
+; CHECK-NEXT:  .LBB4_8:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 12(sp)
+; CHECK-NEXT:    bnez a0, .LBB4_10
+; CHECK-NEXT:  # %bb.9:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 5
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB4_11
+; CHECK-NEXT:  .LBB4_10:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 5
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB4_8
-; CHECK-NEXT:  # %bb.7:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB4_8:
-; CHECK-NEXT:    fsh ft1, 10(sp)
+; CHECK-NEXT:  .LBB4_11:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 10(sp)
+; CHECK-NEXT:    bnez a0, .LBB4_13
+; CHECK-NEXT:  # %bb.12:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 4
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB4_14
+; CHECK-NEXT:  .LBB4_13:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 4
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB4_10
-; CHECK-NEXT:  # %bb.9:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB4_10:
-; CHECK-NEXT:    fsh ft1, 8(sp)
+; CHECK-NEXT:  .LBB4_14:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 8(sp)
+; CHECK-NEXT:    bnez a0, .LBB4_16
+; CHECK-NEXT:  # %bb.15:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 3
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB4_17
+; CHECK-NEXT:  .LBB4_16:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 3
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB4_12
-; CHECK-NEXT:  # %bb.11:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB4_12:
-; CHECK-NEXT:    fsh ft1, 6(sp)
+; CHECK-NEXT:  .LBB4_17:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 6(sp)
+; CHECK-NEXT:    bnez a0, .LBB4_19
+; CHECK-NEXT:  # %bb.18:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 2
-; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 2
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB4_14
-; CHECK-NEXT:  # %bb.13:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB4_14:
-; CHECK-NEXT:    fsh ft1, 4(sp)
+; CHECK-NEXT:    j .LBB4_20
+; CHECK-NEXT:  .LBB4_19:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    vslidedown.vi v25, v8, 2
+; CHECK-NEXT:  .LBB4_20:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 4(sp)
+; CHECK-NEXT:    bnez a0, .LBB4_22
+; CHECK-NEXT:  # %bb.21:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB4_23
+; CHECK-NEXT:  .LBB4_22:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB4_16
-; CHECK-NEXT:  # %bb.15:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB4_16:
-; CHECK-NEXT:    fsh ft1, 2(sp)
+; CHECK-NEXT:  .LBB4_23:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 2(sp)
 ; CHECK-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
 ; CHECK-NEXT:    vle16.v v8, (sp)
 ; CHECK-NEXT:    addi sp, sp, 16
@@ -260,84 +285,93 @@
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    feq.h a0, fa0, fa1
-; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB5_2
+; CHECK-NEXT:    bnez a0, .LBB5_3
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.h ft0, ft1
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    fsh ft0, 0(sp)
+; CHECK-NEXT:    beqz a0, .LBB5_4
 ; CHECK-NEXT:  .LBB5_2:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 7
+; CHECK-NEXT:    j .LBB5_5
+; CHECK-NEXT:  .LBB5_3:
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fsh ft0, 0(sp)
+; CHECK-NEXT:    bnez a0, .LBB5_2
+; CHECK-NEXT:  .LBB5_4:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 7
+; CHECK-NEXT:  .LBB5_5:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 7
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB5_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB5_4:
-; CHECK-NEXT:    fsh ft1, 14(sp)
+; CHECK-NEXT:    fsh ft0, 14(sp)
+; CHECK-NEXT:    bnez a0, .LBB5_7
+; CHECK-NEXT:  # %bb.6:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 6
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB5_8
+; CHECK-NEXT:  .LBB5_7:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 6
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB5_6
-; CHECK-NEXT:  # %bb.5:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB5_6:
-; CHECK-NEXT:    fsh ft1, 12(sp)
+; CHECK-NEXT:  .LBB5_8:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 12(sp)
+; CHECK-NEXT:    bnez a0, .LBB5_10
+; CHECK-NEXT:  # %bb.9:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 5
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB5_11
+; CHECK-NEXT:  .LBB5_10:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 5
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB5_8
-; CHECK-NEXT:  # %bb.7:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB5_8:
-; CHECK-NEXT:    fsh ft1, 10(sp)
+; CHECK-NEXT:  .LBB5_11:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 10(sp)
+; CHECK-NEXT:    bnez a0, .LBB5_13
+; CHECK-NEXT:  # %bb.12:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 4
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB5_14
+; CHECK-NEXT:  .LBB5_13:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 4
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB5_10
-; CHECK-NEXT:  # %bb.9:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB5_10:
-; CHECK-NEXT:    fsh ft1, 8(sp)
+; CHECK-NEXT:  .LBB5_14:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 8(sp)
+; CHECK-NEXT:    bnez a0, .LBB5_16
+; CHECK-NEXT:  # %bb.15:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 3
-; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    j .LBB5_17
+; CHECK-NEXT:  .LBB5_16:
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 3
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB5_12
-; CHECK-NEXT:  # %bb.11:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB5_12:
-; CHECK-NEXT:    fsh ft1, 6(sp)
+; CHECK-NEXT:  .LBB5_17:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 6(sp)
+; CHECK-NEXT:    bnez a0, .LBB5_19
+; CHECK-NEXT:  # %bb.18:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 2
-; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 2
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB5_14
-; CHECK-NEXT:  # %bb.13:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB5_14:
-; CHECK-NEXT:    fsh ft1, 4(sp)
+; CHECK-NEXT:    j .LBB5_20
+; CHECK-NEXT:  .LBB5_19:
 ; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    vslidedown.vi v25, v8, 2
+; CHECK-NEXT:  .LBB5_20:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 4(sp)
+; CHECK-NEXT:    bnez a0, .LBB5_22
+; CHECK-NEXT:  # %bb.21:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB5_23
+; CHECK-NEXT:  .LBB5_22:
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB5_16
-; CHECK-NEXT:  # %bb.15:
-; CHECK-NEXT:    fmv.h ft1, ft0
-; CHECK-NEXT:  .LBB5_16:
-; CHECK-NEXT:    fsh ft1, 2(sp)
+; CHECK-NEXT:  .LBB5_23:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsh ft0, 2(sp)
 ; CHECK-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
 ; CHECK-NEXT:    vle16.v v8, (sp)
 ; CHECK-NEXT:    addi sp, sp, 16
@@ -359,165 +393,182 @@
 ; RV32-NEXT:    addi s0, sp, 64
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -32
-; RV32-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v10
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB6_2
+; RV32-NEXT:    bnez a0, .LBB6_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.h ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v10
+; RV32-NEXT:    fsh ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB6_4
 ; RV32-NEXT:  .LBB6_2:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 15
+; RV32-NEXT:    j .LBB6_5
+; RV32-NEXT:  .LBB6_3:
+; RV32-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsh ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB6_2
+; RV32-NEXT:  .LBB6_4:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 15
+; RV32-NEXT:  .LBB6_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 15
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_4:
-; RV32-NEXT:    fsh ft1, 30(sp)
+; RV32-NEXT:    fsh ft0, 30(sp)
+; RV32-NEXT:    bnez a0, .LBB6_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 14
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 14
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_6:
-; RV32-NEXT:    fsh ft1, 28(sp)
+; RV32-NEXT:    j .LBB6_8
+; RV32-NEXT:  .LBB6_7:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 13
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 13
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.h ft1, ft0
+; RV32-NEXT:    vslidedown.vi v26, v8, 14
 ; RV32-NEXT:  .LBB6_8:
-; RV32-NEXT:    fsh ft1, 26(sp)
-; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 12
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 12
-; RV32-NEXT:    vfmv.f.s ft1, v26
+; RV32-NEXT:    fsh ft0, 28(sp)
 ; RV32-NEXT:    bnez a0, .LBB6_10
 ; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.h ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 13
+; RV32-NEXT:    j .LBB6_11
 ; RV32-NEXT:  .LBB6_10:
-; RV32-NEXT:    fsh ft1, 24(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 11
+; RV32-NEXT:    vslidedown.vi v26, v8, 13
+; RV32-NEXT:  .LBB6_11:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 11
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_12:
-; RV32-NEXT:    fsh ft1, 22(sp)
+; RV32-NEXT:    fsh ft0, 26(sp)
+; RV32-NEXT:    bnez a0, .LBB6_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 10
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 10
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_14:
-; RV32-NEXT:    fsh ft1, 20(sp)
+; RV32-NEXT:    vslidedown.vi v26, v10, 12
+; RV32-NEXT:    j .LBB6_14
+; RV32-NEXT:  .LBB6_13:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 9
+; RV32-NEXT:    vslidedown.vi v26, v8, 12
+; RV32-NEXT:  .LBB6_14:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 9
-; RV32-NEXT:    vfmv.f.s ft1, v26
+; RV32-NEXT:    fsh ft0, 24(sp)
 ; RV32-NEXT:    bnez a0, .LBB6_16
 ; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.h ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 11
+; RV32-NEXT:    j .LBB6_17
 ; RV32-NEXT:  .LBB6_16:
-; RV32-NEXT:    fsh ft1, 18(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 8
+; RV32-NEXT:    vslidedown.vi v26, v8, 11
+; RV32-NEXT:  .LBB6_17:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 8
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_18
-; RV32-NEXT:  # %bb.17:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_18:
-; RV32-NEXT:    fsh ft1, 16(sp)
+; RV32-NEXT:    fsh ft0, 22(sp)
+; RV32-NEXT:    bnez a0, .LBB6_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 7
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_20
-; RV32-NEXT:  # %bb.19:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_20:
-; RV32-NEXT:    fsh ft1, 14(sp)
+; RV32-NEXT:    vslidedown.vi v26, v10, 10
+; RV32-NEXT:    j .LBB6_20
+; RV32-NEXT:  .LBB6_19:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 6
+; RV32-NEXT:    vslidedown.vi v26, v8, 10
+; RV32-NEXT:  .LBB6_20:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v26
+; RV32-NEXT:    fsh ft0, 20(sp)
 ; RV32-NEXT:    bnez a0, .LBB6_22
 ; RV32-NEXT:  # %bb.21:
-; RV32-NEXT:    fmv.h ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 9
+; RV32-NEXT:    j .LBB6_23
 ; RV32-NEXT:  .LBB6_22:
-; RV32-NEXT:    fsh ft1, 12(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 5
+; RV32-NEXT:    vslidedown.vi v26, v8, 9
+; RV32-NEXT:  .LBB6_23:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_24
-; RV32-NEXT:  # %bb.23:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_24:
-; RV32-NEXT:    fsh ft1, 10(sp)
+; RV32-NEXT:    fsh ft0, 18(sp)
+; RV32-NEXT:    bnez a0, .LBB6_25
+; RV32-NEXT:  # %bb.24:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 4
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_26
-; RV32-NEXT:  # %bb.25:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_26:
-; RV32-NEXT:    fsh ft1, 8(sp)
+; RV32-NEXT:    vslidedown.vi v26, v10, 8
+; RV32-NEXT:    j .LBB6_26
+; RV32-NEXT:  .LBB6_25:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 3
+; RV32-NEXT:    vslidedown.vi v26, v8, 8
+; RV32-NEXT:  .LBB6_26:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v26
+; RV32-NEXT:    fsh ft0, 16(sp)
 ; RV32-NEXT:    bnez a0, .LBB6_28
 ; RV32-NEXT:  # %bb.27:
-; RV32-NEXT:    fmv.h ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 7
+; RV32-NEXT:    j .LBB6_29
 ; RV32-NEXT:  .LBB6_28:
-; RV32-NEXT:    fsh ft1, 6(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 2
+; RV32-NEXT:    vslidedown.vi v26, v8, 7
+; RV32-NEXT:  .LBB6_29:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_30
-; RV32-NEXT:  # %bb.29:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB6_30:
-; RV32-NEXT:    fsh ft1, 4(sp)
+; RV32-NEXT:    fsh ft0, 14(sp)
+; RV32-NEXT:    bnez a0, .LBB6_31
+; RV32-NEXT:  # %bb.30:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 1
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB6_32
-; RV32-NEXT:  # %bb.31:
-; RV32-NEXT:    fmv.h ft1, ft0
+; RV32-NEXT:    vslidedown.vi v26, v10, 6
+; RV32-NEXT:    j .LBB6_32
+; RV32-NEXT:  .LBB6_31:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 6
 ; RV32-NEXT:  .LBB6_32:
-; RV32-NEXT:    fsh ft1, 2(sp)
-; RV32-NEXT:    vsetivli a0, 16, e16,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 12(sp)
+; RV32-NEXT:    bnez a0, .LBB6_34
+; RV32-NEXT:  # %bb.33:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 5
+; RV32-NEXT:    j .LBB6_35
+; RV32-NEXT:  .LBB6_34:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 5
+; RV32-NEXT:  .LBB6_35:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 10(sp)
+; RV32-NEXT:    bnez a0, .LBB6_37
+; RV32-NEXT:  # %bb.36:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 4
+; RV32-NEXT:    j .LBB6_38
+; RV32-NEXT:  .LBB6_37:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 4
+; RV32-NEXT:  .LBB6_38:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 8(sp)
+; RV32-NEXT:    bnez a0, .LBB6_40
+; RV32-NEXT:  # %bb.39:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 3
+; RV32-NEXT:    j .LBB6_41
+; RV32-NEXT:  .LBB6_40:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 3
+; RV32-NEXT:  .LBB6_41:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 6(sp)
+; RV32-NEXT:    bnez a0, .LBB6_43
+; RV32-NEXT:  # %bb.42:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 2
+; RV32-NEXT:    j .LBB6_44
+; RV32-NEXT:  .LBB6_43:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 2
+; RV32-NEXT:  .LBB6_44:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 4(sp)
+; RV32-NEXT:    bnez a0, .LBB6_46
+; RV32-NEXT:  # %bb.45:
+; RV32-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    j .LBB6_47
+; RV32-NEXT:  .LBB6_46:
+; RV32-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 1
+; RV32-NEXT:  .LBB6_47:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 2(sp)
+; RV32-NEXT:    vsetivli a0, 16, e16,m2,ta,mu
 ; RV32-NEXT:    vle16.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -64
 ; RV32-NEXT:    lw s0, 56(sp) # 4-byte Folded Reload
@@ -536,164 +587,181 @@
 ; RV64-NEXT:    addi s0, sp, 64
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -32
-; RV64-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v10
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB6_2
+; RV64-NEXT:    bnez a0, .LBB6_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.h ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v10
+; RV64-NEXT:    fsh ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB6_4
 ; RV64-NEXT:  .LBB6_2:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 15
+; RV64-NEXT:    j .LBB6_5
+; RV64-NEXT:  .LBB6_3:
+; RV64-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsh ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB6_2
+; RV64-NEXT:  .LBB6_4:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 15
+; RV64-NEXT:  .LBB6_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 15
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_4:
-; RV64-NEXT:    fsh ft1, 30(sp)
+; RV64-NEXT:    fsh ft0, 30(sp)
+; RV64-NEXT:    bnez a0, .LBB6_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 14
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_8
+; RV64-NEXT:  .LBB6_7:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 14
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_6:
-; RV64-NEXT:    fsh ft1, 28(sp)
+; RV64-NEXT:  .LBB6_8:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 28(sp)
+; RV64-NEXT:    bnez a0, .LBB6_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 13
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_11
+; RV64-NEXT:  .LBB6_10:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 13
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_8:
-; RV64-NEXT:    fsh ft1, 26(sp)
+; RV64-NEXT:  .LBB6_11:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 26(sp)
+; RV64-NEXT:    bnez a0, .LBB6_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 12
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_14
+; RV64-NEXT:  .LBB6_13:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 12
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_10:
-; RV64-NEXT:    fsh ft1, 24(sp)
+; RV64-NEXT:  .LBB6_14:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB6_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 11
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_17
+; RV64-NEXT:  .LBB6_16:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 11
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_12:
-; RV64-NEXT:    fsh ft1, 22(sp)
+; RV64-NEXT:  .LBB6_17:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 22(sp)
+; RV64-NEXT:    bnez a0, .LBB6_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 10
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_20
+; RV64-NEXT:  .LBB6_19:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 10
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_14:
-; RV64-NEXT:    fsh ft1, 20(sp)
+; RV64-NEXT:  .LBB6_20:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 20(sp)
+; RV64-NEXT:    bnez a0, .LBB6_22
+; RV64-NEXT:  # %bb.21:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 9
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_23
+; RV64-NEXT:  .LBB6_22:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 9
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_16:
-; RV64-NEXT:    fsh ft1, 18(sp)
+; RV64-NEXT:  .LBB6_23:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 18(sp)
+; RV64-NEXT:    bnez a0, .LBB6_25
+; RV64-NEXT:  # %bb.24:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 8
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_26
+; RV64-NEXT:  .LBB6_25:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 8
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_18
-; RV64-NEXT:  # %bb.17:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_18:
-; RV64-NEXT:    fsh ft1, 16(sp)
+; RV64-NEXT:  .LBB6_26:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB6_28
+; RV64-NEXT:  # %bb.27:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 7
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_29
+; RV64-NEXT:  .LBB6_28:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_20
-; RV64-NEXT:  # %bb.19:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_20:
-; RV64-NEXT:    fsh ft1, 14(sp)
+; RV64-NEXT:  .LBB6_29:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 14(sp)
+; RV64-NEXT:    bnez a0, .LBB6_31
+; RV64-NEXT:  # %bb.30:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 6
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_32
+; RV64-NEXT:  .LBB6_31:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_22
-; RV64-NEXT:  # %bb.21:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_22:
-; RV64-NEXT:    fsh ft1, 12(sp)
+; RV64-NEXT:  .LBB6_32:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 12(sp)
+; RV64-NEXT:    bnez a0, .LBB6_34
+; RV64-NEXT:  # %bb.33:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 5
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_35
+; RV64-NEXT:  .LBB6_34:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_24
-; RV64-NEXT:  # %bb.23:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_24:
-; RV64-NEXT:    fsh ft1, 10(sp)
+; RV64-NEXT:  .LBB6_35:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 10(sp)
+; RV64-NEXT:    bnez a0, .LBB6_37
+; RV64-NEXT:  # %bb.36:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 4
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_38
+; RV64-NEXT:  .LBB6_37:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_26
-; RV64-NEXT:  # %bb.25:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_26:
-; RV64-NEXT:    fsh ft1, 8(sp)
+; RV64-NEXT:  .LBB6_38:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 8(sp)
+; RV64-NEXT:    bnez a0, .LBB6_40
+; RV64-NEXT:  # %bb.39:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 3
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB6_41
+; RV64-NEXT:  .LBB6_40:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_28
-; RV64-NEXT:  # %bb.27:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_28:
-; RV64-NEXT:    fsh ft1, 6(sp)
+; RV64-NEXT:  .LBB6_41:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 6(sp)
+; RV64-NEXT:    bnez a0, .LBB6_43
+; RV64-NEXT:  # %bb.42:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 2
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_30
-; RV64-NEXT:  # %bb.29:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_30:
-; RV64-NEXT:    fsh ft1, 4(sp)
+; RV64-NEXT:    j .LBB6_44
+; RV64-NEXT:  .LBB6_43:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    vslidedown.vi v26, v8, 2
+; RV64-NEXT:  .LBB6_44:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 4(sp)
+; RV64-NEXT:    bnez a0, .LBB6_46
+; RV64-NEXT:  # %bb.45:
+; RV64-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    j .LBB6_47
+; RV64-NEXT:  .LBB6_46:
+; RV64-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB6_32
-; RV64-NEXT:  # %bb.31:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB6_32:
-; RV64-NEXT:    fsh ft1, 2(sp)
+; RV64-NEXT:  .LBB6_47:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 2(sp)
 ; RV64-NEXT:    vsetivli a0, 16, e16,m2,ta,mu
 ; RV64-NEXT:    vle16.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -64
@@ -718,164 +786,181 @@
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -32
 ; RV32-NEXT:    feq.h a0, fa0, fa1
-; RV32-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v10
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB7_2
+; RV32-NEXT:    bnez a0, .LBB7_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.h ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v10
+; RV32-NEXT:    fsh ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB7_4
 ; RV32-NEXT:  .LBB7_2:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 15
+; RV32-NEXT:    j .LBB7_5
+; RV32-NEXT:  .LBB7_3:
+; RV32-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsh ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB7_2
+; RV32-NEXT:  .LBB7_4:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 15
+; RV32-NEXT:  .LBB7_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 15
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_4:
-; RV32-NEXT:    fsh ft1, 30(sp)
+; RV32-NEXT:    fsh ft0, 30(sp)
+; RV32-NEXT:    bnez a0, .LBB7_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 14
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_8
+; RV32-NEXT:  .LBB7_7:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 14
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_6:
-; RV32-NEXT:    fsh ft1, 28(sp)
+; RV32-NEXT:  .LBB7_8:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 28(sp)
+; RV32-NEXT:    bnez a0, .LBB7_10
+; RV32-NEXT:  # %bb.9:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 13
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_11
+; RV32-NEXT:  .LBB7_10:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 13
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_8:
-; RV32-NEXT:    fsh ft1, 26(sp)
+; RV32-NEXT:  .LBB7_11:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 26(sp)
+; RV32-NEXT:    bnez a0, .LBB7_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 12
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_14
+; RV32-NEXT:  .LBB7_13:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 12
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_10
-; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_10:
-; RV32-NEXT:    fsh ft1, 24(sp)
+; RV32-NEXT:  .LBB7_14:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB7_16
+; RV32-NEXT:  # %bb.15:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 11
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_17
+; RV32-NEXT:  .LBB7_16:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 11
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_12:
-; RV32-NEXT:    fsh ft1, 22(sp)
+; RV32-NEXT:  .LBB7_17:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 22(sp)
+; RV32-NEXT:    bnez a0, .LBB7_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 10
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_20
+; RV32-NEXT:  .LBB7_19:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 10
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_14:
-; RV32-NEXT:    fsh ft1, 20(sp)
+; RV32-NEXT:  .LBB7_20:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 20(sp)
+; RV32-NEXT:    bnez a0, .LBB7_22
+; RV32-NEXT:  # %bb.21:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 9
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_23
+; RV32-NEXT:  .LBB7_22:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 9
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_16
-; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_16:
-; RV32-NEXT:    fsh ft1, 18(sp)
+; RV32-NEXT:  .LBB7_23:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 18(sp)
+; RV32-NEXT:    bnez a0, .LBB7_25
+; RV32-NEXT:  # %bb.24:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 8
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_26
+; RV32-NEXT:  .LBB7_25:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 8
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_18
-; RV32-NEXT:  # %bb.17:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_18:
-; RV32-NEXT:    fsh ft1, 16(sp)
+; RV32-NEXT:  .LBB7_26:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB7_28
+; RV32-NEXT:  # %bb.27:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 7
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_29
+; RV32-NEXT:  .LBB7_28:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_20
-; RV32-NEXT:  # %bb.19:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_20:
-; RV32-NEXT:    fsh ft1, 14(sp)
+; RV32-NEXT:  .LBB7_29:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 14(sp)
+; RV32-NEXT:    bnez a0, .LBB7_31
+; RV32-NEXT:  # %bb.30:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 6
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_32
+; RV32-NEXT:  .LBB7_31:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_22
-; RV32-NEXT:  # %bb.21:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_22:
-; RV32-NEXT:    fsh ft1, 12(sp)
+; RV32-NEXT:  .LBB7_32:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 12(sp)
+; RV32-NEXT:    bnez a0, .LBB7_34
+; RV32-NEXT:  # %bb.33:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 5
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_35
+; RV32-NEXT:  .LBB7_34:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_24
-; RV32-NEXT:  # %bb.23:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_24:
-; RV32-NEXT:    fsh ft1, 10(sp)
+; RV32-NEXT:  .LBB7_35:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 10(sp)
+; RV32-NEXT:    bnez a0, .LBB7_37
+; RV32-NEXT:  # %bb.36:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 4
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_38
+; RV32-NEXT:  .LBB7_37:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_26
-; RV32-NEXT:  # %bb.25:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_26:
-; RV32-NEXT:    fsh ft1, 8(sp)
+; RV32-NEXT:  .LBB7_38:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 8(sp)
+; RV32-NEXT:    bnez a0, .LBB7_40
+; RV32-NEXT:  # %bb.39:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 3
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB7_41
+; RV32-NEXT:  .LBB7_40:
+; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_28
-; RV32-NEXT:  # %bb.27:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_28:
-; RV32-NEXT:    fsh ft1, 6(sp)
+; RV32-NEXT:  .LBB7_41:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 6(sp)
+; RV32-NEXT:    bnez a0, .LBB7_43
+; RV32-NEXT:  # %bb.42:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 2
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_30
-; RV32-NEXT:  # %bb.29:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_30:
-; RV32-NEXT:    fsh ft1, 4(sp)
+; RV32-NEXT:    j .LBB7_44
+; RV32-NEXT:  .LBB7_43:
 ; RV32-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    vslidedown.vi v26, v8, 2
+; RV32-NEXT:  .LBB7_44:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 4(sp)
+; RV32-NEXT:    bnez a0, .LBB7_46
+; RV32-NEXT:  # %bb.45:
+; RV32-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    j .LBB7_47
+; RV32-NEXT:  .LBB7_46:
+; RV32-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB7_32
-; RV32-NEXT:  # %bb.31:
-; RV32-NEXT:    fmv.h ft1, ft0
-; RV32-NEXT:  .LBB7_32:
-; RV32-NEXT:    fsh ft1, 2(sp)
+; RV32-NEXT:  .LBB7_47:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsh ft0, 2(sp)
 ; RV32-NEXT:    vsetivli a0, 16, e16,m2,ta,mu
 ; RV32-NEXT:    vle16.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -64
@@ -896,164 +981,181 @@
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -32
 ; RV64-NEXT:    feq.h a0, fa0, fa1
-; RV64-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v10
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB7_2
+; RV64-NEXT:    bnez a0, .LBB7_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.h ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v10
+; RV64-NEXT:    fsh ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB7_4
 ; RV64-NEXT:  .LBB7_2:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 15
+; RV64-NEXT:    j .LBB7_5
+; RV64-NEXT:  .LBB7_3:
+; RV64-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsh ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB7_2
+; RV64-NEXT:  .LBB7_4:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 15
+; RV64-NEXT:  .LBB7_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 15
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_4:
-; RV64-NEXT:    fsh ft1, 30(sp)
+; RV64-NEXT:    fsh ft0, 30(sp)
+; RV64-NEXT:    bnez a0, .LBB7_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 14
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 14
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_6:
-; RV64-NEXT:    fsh ft1, 28(sp)
+; RV64-NEXT:    j .LBB7_8
+; RV64-NEXT:  .LBB7_7:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 13
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 13
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.h ft1, ft0
+; RV64-NEXT:    vslidedown.vi v26, v8, 14
 ; RV64-NEXT:  .LBB7_8:
-; RV64-NEXT:    fsh ft1, 26(sp)
-; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 12
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 12
-; RV64-NEXT:    vfmv.f.s ft1, v26
+; RV64-NEXT:    fsh ft0, 28(sp)
 ; RV64-NEXT:    bnez a0, .LBB7_10
 ; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.h ft1, ft0
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 13
+; RV64-NEXT:    j .LBB7_11
 ; RV64-NEXT:  .LBB7_10:
-; RV64-NEXT:    fsh ft1, 24(sp)
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 11
+; RV64-NEXT:    vslidedown.vi v26, v8, 13
+; RV64-NEXT:  .LBB7_11:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 11
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_12:
-; RV64-NEXT:    fsh ft1, 22(sp)
+; RV64-NEXT:    fsh ft0, 26(sp)
+; RV64-NEXT:    bnez a0, .LBB7_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 10
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 10
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_14:
-; RV64-NEXT:    fsh ft1, 20(sp)
+; RV64-NEXT:    vslidedown.vi v26, v10, 12
+; RV64-NEXT:    j .LBB7_14
+; RV64-NEXT:  .LBB7_13:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 9
+; RV64-NEXT:    vslidedown.vi v26, v8, 12
+; RV64-NEXT:  .LBB7_14:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 9
-; RV64-NEXT:    vfmv.f.s ft1, v26
+; RV64-NEXT:    fsh ft0, 24(sp)
 ; RV64-NEXT:    bnez a0, .LBB7_16
 ; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.h ft1, ft0
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 11
+; RV64-NEXT:    j .LBB7_17
 ; RV64-NEXT:  .LBB7_16:
-; RV64-NEXT:    fsh ft1, 18(sp)
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 8
+; RV64-NEXT:    vslidedown.vi v26, v8, 11
+; RV64-NEXT:  .LBB7_17:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 8
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_18
-; RV64-NEXT:  # %bb.17:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_18:
-; RV64-NEXT:    fsh ft1, 16(sp)
+; RV64-NEXT:    fsh ft0, 22(sp)
+; RV64-NEXT:    bnez a0, .LBB7_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 7
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_20
-; RV64-NEXT:  # %bb.19:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_20:
-; RV64-NEXT:    fsh ft1, 14(sp)
+; RV64-NEXT:    vslidedown.vi v26, v10, 10
+; RV64-NEXT:    j .LBB7_20
+; RV64-NEXT:  .LBB7_19:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 6
+; RV64-NEXT:    vslidedown.vi v26, v8, 10
+; RV64-NEXT:  .LBB7_20:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v26
+; RV64-NEXT:    fsh ft0, 20(sp)
 ; RV64-NEXT:    bnez a0, .LBB7_22
 ; RV64-NEXT:  # %bb.21:
-; RV64-NEXT:    fmv.h ft1, ft0
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 9
+; RV64-NEXT:    j .LBB7_23
 ; RV64-NEXT:  .LBB7_22:
-; RV64-NEXT:    fsh ft1, 12(sp)
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 5
+; RV64-NEXT:    vslidedown.vi v26, v8, 9
+; RV64-NEXT:  .LBB7_23:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 18(sp)
+; RV64-NEXT:    bnez a0, .LBB7_25
+; RV64-NEXT:  # %bb.24:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 8
+; RV64-NEXT:    j .LBB7_26
+; RV64-NEXT:  .LBB7_25:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 8
+; RV64-NEXT:  .LBB7_26:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB7_28
+; RV64-NEXT:  # %bb.27:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 7
+; RV64-NEXT:    j .LBB7_29
+; RV64-NEXT:  .LBB7_28:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 7
+; RV64-NEXT:  .LBB7_29:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 14(sp)
+; RV64-NEXT:    bnez a0, .LBB7_31
+; RV64-NEXT:  # %bb.30:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 6
+; RV64-NEXT:    j .LBB7_32
+; RV64-NEXT:  .LBB7_31:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 6
+; RV64-NEXT:  .LBB7_32:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 12(sp)
+; RV64-NEXT:    bnez a0, .LBB7_34
+; RV64-NEXT:  # %bb.33:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 5
+; RV64-NEXT:    j .LBB7_35
+; RV64-NEXT:  .LBB7_34:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_24
-; RV64-NEXT:  # %bb.23:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_24:
-; RV64-NEXT:    fsh ft1, 10(sp)
+; RV64-NEXT:  .LBB7_35:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 10(sp)
+; RV64-NEXT:    bnez a0, .LBB7_37
+; RV64-NEXT:  # %bb.36:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 4
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB7_38
+; RV64-NEXT:  .LBB7_37:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_26
-; RV64-NEXT:  # %bb.25:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_26:
-; RV64-NEXT:    fsh ft1, 8(sp)
+; RV64-NEXT:  .LBB7_38:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 8(sp)
+; RV64-NEXT:    bnez a0, .LBB7_40
+; RV64-NEXT:  # %bb.39:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 3
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB7_41
+; RV64-NEXT:  .LBB7_40:
+; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_28
-; RV64-NEXT:  # %bb.27:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_28:
-; RV64-NEXT:    fsh ft1, 6(sp)
+; RV64-NEXT:  .LBB7_41:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 6(sp)
+; RV64-NEXT:    bnez a0, .LBB7_43
+; RV64-NEXT:  # %bb.42:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 2
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_30
-; RV64-NEXT:  # %bb.29:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_30:
-; RV64-NEXT:    fsh ft1, 4(sp)
+; RV64-NEXT:    j .LBB7_44
+; RV64-NEXT:  .LBB7_43:
 ; RV64-NEXT:    vsetivli a1, 1, e16,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    vslidedown.vi v26, v8, 2
+; RV64-NEXT:  .LBB7_44:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 4(sp)
+; RV64-NEXT:    bnez a0, .LBB7_46
+; RV64-NEXT:  # %bb.45:
+; RV64-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    j .LBB7_47
+; RV64-NEXT:  .LBB7_46:
+; RV64-NEXT:    vsetivli a0, 1, e16,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB7_32
-; RV64-NEXT:  # %bb.31:
-; RV64-NEXT:    fmv.h ft1, ft0
-; RV64-NEXT:  .LBB7_32:
-; RV64-NEXT:    fsh ft1, 2(sp)
+; RV64-NEXT:  .LBB7_47:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsh ft0, 2(sp)
 ; RV64-NEXT:    vsetivli a0, 16, e16,m2,ta,mu
 ; RV64-NEXT:    vle16.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -64
@@ -1069,20 +1171,22 @@
 define <2 x float> @select_v2f32(i1 zeroext %c, <2 x float> %a, <2 x float> %b) {
 ; CHECK-LABEL: select_v2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
-; CHECK-NEXT:    vfmv.f.s ft3, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft2, v25
 ; CHECK-NEXT:    bnez a0, .LBB8_2
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.s ft0, ft1
-; CHECK-NEXT:    fmv.s ft2, ft3
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    vsetivli a0, 1, e32,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB8_3
 ; CHECK-NEXT:  .LBB8_2:
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
+; CHECK-NEXT:    vsetivli a0, 1, e32,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 1
+; CHECK-NEXT:  .LBB8_3:
+; CHECK-NEXT:    vfmv.f.s ft1, v25
 ; CHECK-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
-; CHECK-NEXT:    vfmv.v.f v8, ft2
+; CHECK-NEXT:    vfmv.v.f v8, ft1
 ; CHECK-NEXT:    vfmv.s.f v8, ft0
 ; CHECK-NEXT:    ret
   %v = select i1 %c, <2 x float> %a, <2 x float> %b
@@ -1093,23 +1197,27 @@
 ; CHECK-LABEL: selectcc_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    feq.s a0, fa0, fa1
+; CHECK-NEXT:    bnez a0, .LBB9_2
+; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
+; CHECK-NEXT:    j .LBB9_3
+; CHECK-NEXT:  .LBB9_2:
+; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
+; CHECK-NEXT:  .LBB9_3:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    bnez a0, .LBB9_2
-; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.s ft0, ft1
-; CHECK-NEXT:  .LBB9_2:
 ; CHECK-NEXT:    vsetivli a1, 2, e32,m1,ta,mu
 ; CHECK-NEXT:    vfmv.v.f v25, ft0
-; CHECK-NEXT:    vfmv.f.s ft1, v9
+; CHECK-NEXT:    bnez a0, .LBB9_5
+; CHECK-NEXT:  # %bb.4:
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    j .LBB9_6
+; CHECK-NEXT:  .LBB9_5:
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB9_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.s ft0, ft1
-; CHECK-NEXT:  .LBB9_4:
+; CHECK-NEXT:  .LBB9_6:
 ; CHECK-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
 ; CHECK-NEXT:    vfmv.s.f v25, ft0
 ; CHECK-NEXT:    vmv1r.v v8, v25
@@ -1124,44 +1232,49 @@
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
-; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB10_2
+; CHECK-NEXT:    bnez a0, .LBB10_3
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.s ft0, ft1
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    fsw ft0, 0(sp)
+; CHECK-NEXT:    beqz a0, .LBB10_4
 ; CHECK-NEXT:  .LBB10_2:
+; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 3
+; CHECK-NEXT:    j .LBB10_5
+; CHECK-NEXT:  .LBB10_3:
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fsw ft0, 0(sp)
+; CHECK-NEXT:    bnez a0, .LBB10_2
+; CHECK-NEXT:  .LBB10_4:
 ; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 3
+; CHECK-NEXT:  .LBB10_5:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 3
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB10_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.s ft1, ft0
-; CHECK-NEXT:  .LBB10_4:
-; CHECK-NEXT:    fsw ft1, 12(sp)
+; CHECK-NEXT:    fsw ft0, 12(sp)
+; CHECK-NEXT:    bnez a0, .LBB10_7
+; CHECK-NEXT:  # %bb.6:
 ; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 2
-; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 2
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB10_6
-; CHECK-NEXT:  # %bb.5:
-; CHECK-NEXT:    fmv.s ft1, ft0
-; CHECK-NEXT:  .LBB10_6:
-; CHECK-NEXT:    fsw ft1, 8(sp)
+; CHECK-NEXT:    j .LBB10_8
+; CHECK-NEXT:  .LBB10_7:
 ; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    vslidedown.vi v25, v8, 2
+; CHECK-NEXT:  .LBB10_8:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsw ft0, 8(sp)
+; CHECK-NEXT:    bnez a0, .LBB10_10
+; CHECK-NEXT:  # %bb.9:
+; CHECK-NEXT:    vsetivli a0, 1, e32,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB10_11
+; CHECK-NEXT:  .LBB10_10:
+; CHECK-NEXT:    vsetivli a0, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB10_8
-; CHECK-NEXT:  # %bb.7:
-; CHECK-NEXT:    fmv.s ft1, ft0
-; CHECK-NEXT:  .LBB10_8:
-; CHECK-NEXT:    fsw ft1, 4(sp)
+; CHECK-NEXT:  .LBB10_11:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsw ft0, 4(sp)
 ; CHECK-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
 ; CHECK-NEXT:    vle32.v v8, (sp)
 ; CHECK-NEXT:    addi sp, sp, 16
@@ -1176,44 +1289,49 @@
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    feq.s a0, fa0, fa1
-; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB11_2
+; CHECK-NEXT:    bnez a0, .LBB11_3
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.s ft0, ft1
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    fsw ft0, 0(sp)
+; CHECK-NEXT:    beqz a0, .LBB11_4
 ; CHECK-NEXT:  .LBB11_2:
+; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 3
+; CHECK-NEXT:    j .LBB11_5
+; CHECK-NEXT:  .LBB11_3:
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fsw ft0, 0(sp)
+; CHECK-NEXT:    bnez a0, .LBB11_2
+; CHECK-NEXT:  .LBB11_4:
 ; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 3
+; CHECK-NEXT:  .LBB11_5:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 3
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB11_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.s ft1, ft0
-; CHECK-NEXT:  .LBB11_4:
-; CHECK-NEXT:    fsw ft1, 12(sp)
+; CHECK-NEXT:    fsw ft0, 12(sp)
+; CHECK-NEXT:    bnez a0, .LBB11_7
+; CHECK-NEXT:  # %bb.6:
 ; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 2
-; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 2
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB11_6
-; CHECK-NEXT:  # %bb.5:
-; CHECK-NEXT:    fmv.s ft1, ft0
-; CHECK-NEXT:  .LBB11_6:
-; CHECK-NEXT:    fsw ft1, 8(sp)
+; CHECK-NEXT:    j .LBB11_8
+; CHECK-NEXT:  .LBB11_7:
 ; CHECK-NEXT:    vsetivli a1, 1, e32,m1,ta,mu
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    vslidedown.vi v25, v8, 2
+; CHECK-NEXT:  .LBB11_8:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsw ft0, 8(sp)
+; CHECK-NEXT:    bnez a0, .LBB11_10
+; CHECK-NEXT:  # %bb.9:
+; CHECK-NEXT:    vsetivli a0, 1, e32,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB11_11
+; CHECK-NEXT:  .LBB11_10:
+; CHECK-NEXT:    vsetivli a0, 1, e32,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
-; CHECK-NEXT:    bnez a0, .LBB11_8
-; CHECK-NEXT:  # %bb.7:
-; CHECK-NEXT:    fmv.s ft1, ft0
-; CHECK-NEXT:  .LBB11_8:
-; CHECK-NEXT:    fsw ft1, 4(sp)
+; CHECK-NEXT:  .LBB11_11:
+; CHECK-NEXT:    vfmv.f.s ft0, v25
+; CHECK-NEXT:    fsw ft0, 4(sp)
 ; CHECK-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
 ; CHECK-NEXT:    vle32.v v8, (sp)
 ; CHECK-NEXT:    addi sp, sp, 16
@@ -1235,84 +1353,93 @@
 ; RV32-NEXT:    addi s0, sp, 64
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -32
-; RV32-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v10
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB12_2
+; RV32-NEXT:    bnez a0, .LBB12_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.s ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v10
+; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB12_4
 ; RV32-NEXT:  .LBB12_2:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 7
+; RV32-NEXT:    j .LBB12_5
+; RV32-NEXT:  .LBB12_3:
+; RV32-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB12_2
+; RV32-NEXT:  .LBB12_4:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 7
+; RV32-NEXT:  .LBB12_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB12_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB12_4:
-; RV32-NEXT:    fsw ft1, 28(sp)
+; RV32-NEXT:    fsw ft0, 28(sp)
+; RV32-NEXT:    bnez a0, .LBB12_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 6
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB12_8
+; RV32-NEXT:  .LBB12_7:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB12_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB12_6:
-; RV32-NEXT:    fsw ft1, 24(sp)
+; RV32-NEXT:  .LBB12_8:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB12_10
+; RV32-NEXT:  # %bb.9:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 5
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB12_11
+; RV32-NEXT:  .LBB12_10:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB12_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB12_8:
-; RV32-NEXT:    fsw ft1, 20(sp)
+; RV32-NEXT:  .LBB12_11:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 20(sp)
+; RV32-NEXT:    bnez a0, .LBB12_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 4
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB12_14
+; RV32-NEXT:  .LBB12_13:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB12_10
-; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB12_10:
-; RV32-NEXT:    fsw ft1, 16(sp)
+; RV32-NEXT:  .LBB12_14:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB12_16
+; RV32-NEXT:  # %bb.15:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 3
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB12_17
+; RV32-NEXT:  .LBB12_16:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB12_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB12_12:
-; RV32-NEXT:    fsw ft1, 12(sp)
+; RV32-NEXT:  .LBB12_17:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 12(sp)
+; RV32-NEXT:    bnez a0, .LBB12_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 2
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB12_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB12_14:
-; RV32-NEXT:    fsw ft1, 8(sp)
+; RV32-NEXT:    j .LBB12_20
+; RV32-NEXT:  .LBB12_19:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    vslidedown.vi v26, v8, 2
+; RV32-NEXT:  .LBB12_20:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 8(sp)
+; RV32-NEXT:    bnez a0, .LBB12_22
+; RV32-NEXT:  # %bb.21:
+; RV32-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    j .LBB12_23
+; RV32-NEXT:  .LBB12_22:
+; RV32-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB12_16
-; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB12_16:
-; RV32-NEXT:    fsw ft1, 4(sp)
+; RV32-NEXT:  .LBB12_23:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 4(sp)
 ; RV32-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
 ; RV32-NEXT:    vle32.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -64
@@ -1332,84 +1459,93 @@
 ; RV64-NEXT:    addi s0, sp, 64
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -32
-; RV64-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v10
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB12_2
+; RV64-NEXT:    bnez a0, .LBB12_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.s ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v10
+; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB12_4
 ; RV64-NEXT:  .LBB12_2:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 7
+; RV64-NEXT:    j .LBB12_5
+; RV64-NEXT:  .LBB12_3:
+; RV64-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB12_2
+; RV64-NEXT:  .LBB12_4:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 7
+; RV64-NEXT:  .LBB12_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB12_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB12_4:
-; RV64-NEXT:    fsw ft1, 28(sp)
+; RV64-NEXT:    fsw ft0, 28(sp)
+; RV64-NEXT:    bnez a0, .LBB12_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 6
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB12_8
+; RV64-NEXT:  .LBB12_7:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB12_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB12_6:
-; RV64-NEXT:    fsw ft1, 24(sp)
+; RV64-NEXT:  .LBB12_8:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB12_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 5
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB12_11
+; RV64-NEXT:  .LBB12_10:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB12_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB12_8:
-; RV64-NEXT:    fsw ft1, 20(sp)
+; RV64-NEXT:  .LBB12_11:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 20(sp)
+; RV64-NEXT:    bnez a0, .LBB12_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 4
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB12_14
+; RV64-NEXT:  .LBB12_13:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB12_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB12_10:
-; RV64-NEXT:    fsw ft1, 16(sp)
+; RV64-NEXT:  .LBB12_14:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB12_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 3
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB12_17
+; RV64-NEXT:  .LBB12_16:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB12_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB12_12:
-; RV64-NEXT:    fsw ft1, 12(sp)
+; RV64-NEXT:  .LBB12_17:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 12(sp)
+; RV64-NEXT:    bnez a0, .LBB12_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 2
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB12_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB12_14:
-; RV64-NEXT:    fsw ft1, 8(sp)
+; RV64-NEXT:    j .LBB12_20
+; RV64-NEXT:  .LBB12_19:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    vslidedown.vi v26, v8, 2
+; RV64-NEXT:  .LBB12_20:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 8(sp)
+; RV64-NEXT:    bnez a0, .LBB12_22
+; RV64-NEXT:  # %bb.21:
+; RV64-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    j .LBB12_23
+; RV64-NEXT:  .LBB12_22:
+; RV64-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB12_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB12_16:
-; RV64-NEXT:    fsw ft1, 4(sp)
+; RV64-NEXT:  .LBB12_23:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 4(sp)
 ; RV64-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
 ; RV64-NEXT:    vle32.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -64
@@ -1434,84 +1570,93 @@
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -32
 ; RV32-NEXT:    feq.s a0, fa0, fa1
-; RV32-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v10
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB13_2
+; RV32-NEXT:    bnez a0, .LBB13_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.s ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v10
+; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB13_4
 ; RV32-NEXT:  .LBB13_2:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 7
+; RV32-NEXT:    j .LBB13_5
+; RV32-NEXT:  .LBB13_3:
+; RV32-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB13_2
+; RV32-NEXT:  .LBB13_4:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 7
+; RV32-NEXT:  .LBB13_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB13_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB13_4:
-; RV32-NEXT:    fsw ft1, 28(sp)
+; RV32-NEXT:    fsw ft0, 28(sp)
+; RV32-NEXT:    bnez a0, .LBB13_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 6
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB13_8
+; RV32-NEXT:  .LBB13_7:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB13_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB13_6:
-; RV32-NEXT:    fsw ft1, 24(sp)
+; RV32-NEXT:  .LBB13_8:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB13_10
+; RV32-NEXT:  # %bb.9:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 5
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB13_11
+; RV32-NEXT:  .LBB13_10:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB13_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB13_8:
-; RV32-NEXT:    fsw ft1, 20(sp)
+; RV32-NEXT:  .LBB13_11:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 20(sp)
+; RV32-NEXT:    bnez a0, .LBB13_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 4
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB13_14
+; RV32-NEXT:  .LBB13_13:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB13_10
-; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB13_10:
-; RV32-NEXT:    fsw ft1, 16(sp)
+; RV32-NEXT:  .LBB13_14:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB13_16
+; RV32-NEXT:  # %bb.15:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 3
-; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    j .LBB13_17
+; RV32-NEXT:  .LBB13_16:
+; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB13_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB13_12:
-; RV32-NEXT:    fsw ft1, 12(sp)
+; RV32-NEXT:  .LBB13_17:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 12(sp)
+; RV32-NEXT:    bnez a0, .LBB13_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 2
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB13_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB13_14:
-; RV32-NEXT:    fsw ft1, 8(sp)
+; RV32-NEXT:    j .LBB13_20
+; RV32-NEXT:  .LBB13_19:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    vslidedown.vi v26, v8, 2
+; RV32-NEXT:  .LBB13_20:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 8(sp)
+; RV32-NEXT:    bnez a0, .LBB13_22
+; RV32-NEXT:  # %bb.21:
+; RV32-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    j .LBB13_23
+; RV32-NEXT:  .LBB13_22:
+; RV32-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB13_16
-; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB13_16:
-; RV32-NEXT:    fsw ft1, 4(sp)
+; RV32-NEXT:  .LBB13_23:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsw ft0, 4(sp)
 ; RV32-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
 ; RV32-NEXT:    vle32.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -64
@@ -1532,84 +1677,93 @@
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -32
 ; RV64-NEXT:    feq.s a0, fa0, fa1
-; RV64-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v10
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB13_2
+; RV64-NEXT:    bnez a0, .LBB13_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.s ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v10
+; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB13_4
 ; RV64-NEXT:  .LBB13_2:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 7
+; RV64-NEXT:    j .LBB13_5
+; RV64-NEXT:  .LBB13_3:
+; RV64-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB13_2
+; RV64-NEXT:  .LBB13_4:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 7
+; RV64-NEXT:  .LBB13_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB13_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB13_4:
-; RV64-NEXT:    fsw ft1, 28(sp)
+; RV64-NEXT:    fsw ft0, 28(sp)
+; RV64-NEXT:    bnez a0, .LBB13_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 6
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB13_8
+; RV64-NEXT:  .LBB13_7:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB13_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB13_6:
-; RV64-NEXT:    fsw ft1, 24(sp)
+; RV64-NEXT:  .LBB13_8:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB13_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 5
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB13_11
+; RV64-NEXT:  .LBB13_10:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB13_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB13_8:
-; RV64-NEXT:    fsw ft1, 20(sp)
+; RV64-NEXT:  .LBB13_11:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 20(sp)
+; RV64-NEXT:    bnez a0, .LBB13_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 4
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB13_14
+; RV64-NEXT:  .LBB13_13:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB13_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB13_10:
-; RV64-NEXT:    fsw ft1, 16(sp)
+; RV64-NEXT:  .LBB13_14:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB13_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 3
-; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    j .LBB13_17
+; RV64-NEXT:  .LBB13_16:
+; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB13_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB13_12:
-; RV64-NEXT:    fsw ft1, 12(sp)
+; RV64-NEXT:  .LBB13_17:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 12(sp)
+; RV64-NEXT:    bnez a0, .LBB13_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 2
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB13_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB13_14:
-; RV64-NEXT:    fsw ft1, 8(sp)
+; RV64-NEXT:    j .LBB13_20
+; RV64-NEXT:  .LBB13_19:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    vslidedown.vi v26, v8, 2
+; RV64-NEXT:  .LBB13_20:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 8(sp)
+; RV64-NEXT:    bnez a0, .LBB13_22
+; RV64-NEXT:  # %bb.21:
+; RV64-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    j .LBB13_23
+; RV64-NEXT:  .LBB13_22:
+; RV64-NEXT:    vsetivli a0, 1, e32,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB13_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB13_16:
-; RV64-NEXT:    fsw ft1, 4(sp)
+; RV64-NEXT:  .LBB13_23:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsw ft0, 4(sp)
 ; RV64-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
 ; RV64-NEXT:    vle32.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -64
@@ -1634,164 +1788,181 @@
 ; RV32-NEXT:    addi s0, sp, 128
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -64
-; RV32-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v12
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB14_2
+; RV32-NEXT:    bnez a0, .LBB14_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.s ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v12
+; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB14_4
 ; RV32-NEXT:  .LBB14_2:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v8, 15
+; RV32-NEXT:    j .LBB14_5
+; RV32-NEXT:  .LBB14_3:
+; RV32-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB14_2
+; RV32-NEXT:  .LBB14_4:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 15
+; RV32-NEXT:  .LBB14_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 15
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_4:
-; RV32-NEXT:    fsw ft1, 60(sp)
+; RV32-NEXT:    fsw ft0, 60(sp)
+; RV32-NEXT:    bnez a0, .LBB14_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 14
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_8
+; RV32-NEXT:  .LBB14_7:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 14
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_6:
-; RV32-NEXT:    fsw ft1, 56(sp)
+; RV32-NEXT:  .LBB14_8:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 56(sp)
+; RV32-NEXT:    bnez a0, .LBB14_10
+; RV32-NEXT:  # %bb.9:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 13
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_11
+; RV32-NEXT:  .LBB14_10:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 13
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_8:
-; RV32-NEXT:    fsw ft1, 52(sp)
+; RV32-NEXT:  .LBB14_11:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 52(sp)
+; RV32-NEXT:    bnez a0, .LBB14_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 12
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_14
+; RV32-NEXT:  .LBB14_13:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 12
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_10
-; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_10:
-; RV32-NEXT:    fsw ft1, 48(sp)
+; RV32-NEXT:  .LBB14_14:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 48(sp)
+; RV32-NEXT:    bnez a0, .LBB14_16
+; RV32-NEXT:  # %bb.15:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 11
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_17
+; RV32-NEXT:  .LBB14_16:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 11
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_12:
-; RV32-NEXT:    fsw ft1, 44(sp)
+; RV32-NEXT:  .LBB14_17:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 44(sp)
+; RV32-NEXT:    bnez a0, .LBB14_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 10
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_20
+; RV32-NEXT:  .LBB14_19:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 10
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_14:
-; RV32-NEXT:    fsw ft1, 40(sp)
+; RV32-NEXT:  .LBB14_20:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 40(sp)
+; RV32-NEXT:    bnez a0, .LBB14_22
+; RV32-NEXT:  # %bb.21:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 9
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_23
+; RV32-NEXT:  .LBB14_22:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 9
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_16
-; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_16:
-; RV32-NEXT:    fsw ft1, 36(sp)
+; RV32-NEXT:  .LBB14_23:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 36(sp)
+; RV32-NEXT:    bnez a0, .LBB14_25
+; RV32-NEXT:  # %bb.24:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 8
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_26
+; RV32-NEXT:  .LBB14_25:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 8
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_18
-; RV32-NEXT:  # %bb.17:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_18:
-; RV32-NEXT:    fsw ft1, 32(sp)
+; RV32-NEXT:  .LBB14_26:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 32(sp)
+; RV32-NEXT:    bnez a0, .LBB14_28
+; RV32-NEXT:  # %bb.27:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 7
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_29
+; RV32-NEXT:  .LBB14_28:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_20
-; RV32-NEXT:  # %bb.19:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_20:
-; RV32-NEXT:    fsw ft1, 28(sp)
+; RV32-NEXT:  .LBB14_29:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 28(sp)
+; RV32-NEXT:    bnez a0, .LBB14_31
+; RV32-NEXT:  # %bb.30:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 6
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_32
+; RV32-NEXT:  .LBB14_31:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_22
-; RV32-NEXT:  # %bb.21:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_22:
-; RV32-NEXT:    fsw ft1, 24(sp)
+; RV32-NEXT:  .LBB14_32:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB14_34
+; RV32-NEXT:  # %bb.33:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 5
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_35
+; RV32-NEXT:  .LBB14_34:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_24
-; RV32-NEXT:  # %bb.23:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_24:
-; RV32-NEXT:    fsw ft1, 20(sp)
+; RV32-NEXT:  .LBB14_35:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 20(sp)
+; RV32-NEXT:    bnez a0, .LBB14_37
+; RV32-NEXT:  # %bb.36:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 4
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_38
+; RV32-NEXT:  .LBB14_37:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_26
-; RV32-NEXT:  # %bb.25:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_26:
-; RV32-NEXT:    fsw ft1, 16(sp)
+; RV32-NEXT:  .LBB14_38:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB14_40
+; RV32-NEXT:  # %bb.39:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 3
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB14_41
+; RV32-NEXT:  .LBB14_40:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_28
-; RV32-NEXT:  # %bb.27:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_28:
-; RV32-NEXT:    fsw ft1, 12(sp)
+; RV32-NEXT:  .LBB14_41:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 12(sp)
+; RV32-NEXT:    bnez a0, .LBB14_43
+; RV32-NEXT:  # %bb.42:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 2
-; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_30
-; RV32-NEXT:  # %bb.29:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_30:
-; RV32-NEXT:    fsw ft1, 8(sp)
+; RV32-NEXT:    j .LBB14_44
+; RV32-NEXT:  .LBB14_43:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    vslidedown.vi v28, v8, 2
+; RV32-NEXT:  .LBB14_44:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 8(sp)
+; RV32-NEXT:    bnez a0, .LBB14_46
+; RV32-NEXT:  # %bb.45:
+; RV32-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    j .LBB14_47
+; RV32-NEXT:  .LBB14_46:
+; RV32-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB14_32
-; RV32-NEXT:  # %bb.31:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB14_32:
-; RV32-NEXT:    fsw ft1, 4(sp)
+; RV32-NEXT:  .LBB14_47:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 4(sp)
 ; RV32-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
 ; RV32-NEXT:    vle32.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -128
@@ -1811,164 +1982,181 @@
 ; RV64-NEXT:    addi s0, sp, 128
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -64
-; RV64-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v12
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB14_2
+; RV64-NEXT:    bnez a0, .LBB14_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.s ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v12
+; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB14_4
 ; RV64-NEXT:  .LBB14_2:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v8, 15
+; RV64-NEXT:    j .LBB14_5
+; RV64-NEXT:  .LBB14_3:
+; RV64-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB14_2
+; RV64-NEXT:  .LBB14_4:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 15
+; RV64-NEXT:  .LBB14_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 15
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_4:
-; RV64-NEXT:    fsw ft1, 60(sp)
+; RV64-NEXT:    fsw ft0, 60(sp)
+; RV64-NEXT:    bnez a0, .LBB14_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 14
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_8
+; RV64-NEXT:  .LBB14_7:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 14
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_6:
-; RV64-NEXT:    fsw ft1, 56(sp)
+; RV64-NEXT:  .LBB14_8:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 56(sp)
+; RV64-NEXT:    bnez a0, .LBB14_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 13
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_11
+; RV64-NEXT:  .LBB14_10:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 13
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_8:
-; RV64-NEXT:    fsw ft1, 52(sp)
+; RV64-NEXT:  .LBB14_11:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 52(sp)
+; RV64-NEXT:    bnez a0, .LBB14_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 12
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_14
+; RV64-NEXT:  .LBB14_13:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 12
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_10:
-; RV64-NEXT:    fsw ft1, 48(sp)
+; RV64-NEXT:  .LBB14_14:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 48(sp)
+; RV64-NEXT:    bnez a0, .LBB14_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 11
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_17
+; RV64-NEXT:  .LBB14_16:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 11
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_12:
-; RV64-NEXT:    fsw ft1, 44(sp)
+; RV64-NEXT:  .LBB14_17:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 44(sp)
+; RV64-NEXT:    bnez a0, .LBB14_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 10
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_20
+; RV64-NEXT:  .LBB14_19:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 10
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_14:
-; RV64-NEXT:    fsw ft1, 40(sp)
+; RV64-NEXT:  .LBB14_20:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 40(sp)
+; RV64-NEXT:    bnez a0, .LBB14_22
+; RV64-NEXT:  # %bb.21:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 9
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_23
+; RV64-NEXT:  .LBB14_22:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 9
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_16:
-; RV64-NEXT:    fsw ft1, 36(sp)
+; RV64-NEXT:  .LBB14_23:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 36(sp)
+; RV64-NEXT:    bnez a0, .LBB14_25
+; RV64-NEXT:  # %bb.24:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 8
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_26
+; RV64-NEXT:  .LBB14_25:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 8
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_18
-; RV64-NEXT:  # %bb.17:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_18:
-; RV64-NEXT:    fsw ft1, 32(sp)
+; RV64-NEXT:  .LBB14_26:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 32(sp)
+; RV64-NEXT:    bnez a0, .LBB14_28
+; RV64-NEXT:  # %bb.27:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 7
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_29
+; RV64-NEXT:  .LBB14_28:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_20
-; RV64-NEXT:  # %bb.19:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_20:
-; RV64-NEXT:    fsw ft1, 28(sp)
+; RV64-NEXT:  .LBB14_29:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 28(sp)
+; RV64-NEXT:    bnez a0, .LBB14_31
+; RV64-NEXT:  # %bb.30:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 6
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_32
+; RV64-NEXT:  .LBB14_31:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_22
-; RV64-NEXT:  # %bb.21:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_22:
-; RV64-NEXT:    fsw ft1, 24(sp)
+; RV64-NEXT:  .LBB14_32:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB14_34
+; RV64-NEXT:  # %bb.33:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 5
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_35
+; RV64-NEXT:  .LBB14_34:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_24
-; RV64-NEXT:  # %bb.23:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_24:
-; RV64-NEXT:    fsw ft1, 20(sp)
+; RV64-NEXT:  .LBB14_35:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 20(sp)
+; RV64-NEXT:    bnez a0, .LBB14_37
+; RV64-NEXT:  # %bb.36:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 4
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_38
+; RV64-NEXT:  .LBB14_37:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_26
-; RV64-NEXT:  # %bb.25:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_26:
-; RV64-NEXT:    fsw ft1, 16(sp)
+; RV64-NEXT:  .LBB14_38:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB14_40
+; RV64-NEXT:  # %bb.39:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 3
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB14_41
+; RV64-NEXT:  .LBB14_40:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_28
-; RV64-NEXT:  # %bb.27:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_28:
-; RV64-NEXT:    fsw ft1, 12(sp)
+; RV64-NEXT:  .LBB14_41:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 12(sp)
+; RV64-NEXT:    bnez a0, .LBB14_43
+; RV64-NEXT:  # %bb.42:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 2
-; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_30
-; RV64-NEXT:  # %bb.29:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_30:
-; RV64-NEXT:    fsw ft1, 8(sp)
+; RV64-NEXT:    j .LBB14_44
+; RV64-NEXT:  .LBB14_43:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    vslidedown.vi v28, v8, 2
+; RV64-NEXT:  .LBB14_44:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 8(sp)
+; RV64-NEXT:    bnez a0, .LBB14_46
+; RV64-NEXT:  # %bb.45:
+; RV64-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    j .LBB14_47
+; RV64-NEXT:  .LBB14_46:
+; RV64-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB14_32
-; RV64-NEXT:  # %bb.31:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB14_32:
-; RV64-NEXT:    fsw ft1, 4(sp)
+; RV64-NEXT:  .LBB14_47:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 4(sp)
 ; RV64-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
 ; RV64-NEXT:    vle32.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -128
@@ -1993,164 +2181,181 @@
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -64
 ; RV32-NEXT:    feq.s a0, fa0, fa1
-; RV32-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v12
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB15_2
+; RV32-NEXT:    bnez a0, .LBB15_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.s ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v12
+; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB15_4
 ; RV32-NEXT:  .LBB15_2:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v8, 15
+; RV32-NEXT:    j .LBB15_5
+; RV32-NEXT:  .LBB15_3:
+; RV32-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsw ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB15_2
+; RV32-NEXT:  .LBB15_4:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 15
+; RV32-NEXT:  .LBB15_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 15
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_4:
-; RV32-NEXT:    fsw ft1, 60(sp)
+; RV32-NEXT:    fsw ft0, 60(sp)
+; RV32-NEXT:    bnez a0, .LBB15_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 14
-; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 14
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_6:
-; RV32-NEXT:    fsw ft1, 56(sp)
+; RV32-NEXT:    j .LBB15_8
+; RV32-NEXT:  .LBB15_7:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 13
-; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 13
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.s ft1, ft0
+; RV32-NEXT:    vslidedown.vi v28, v8, 14
 ; RV32-NEXT:  .LBB15_8:
-; RV32-NEXT:    fsw ft1, 52(sp)
-; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 12
 ; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 12
-; RV32-NEXT:    vfmv.f.s ft1, v28
+; RV32-NEXT:    fsw ft0, 56(sp)
 ; RV32-NEXT:    bnez a0, .LBB15_10
 ; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.s ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 13
+; RV32-NEXT:    j .LBB15_11
 ; RV32-NEXT:  .LBB15_10:
-; RV32-NEXT:    fsw ft1, 48(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 11
+; RV32-NEXT:    vslidedown.vi v28, v8, 13
+; RV32-NEXT:  .LBB15_11:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 11
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_12:
-; RV32-NEXT:    fsw ft1, 44(sp)
+; RV32-NEXT:    fsw ft0, 52(sp)
+; RV32-NEXT:    bnez a0, .LBB15_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 10
-; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 10
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_14:
-; RV32-NEXT:    fsw ft1, 40(sp)
+; RV32-NEXT:    vslidedown.vi v28, v12, 12
+; RV32-NEXT:    j .LBB15_14
+; RV32-NEXT:  .LBB15_13:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 9
+; RV32-NEXT:    vslidedown.vi v28, v8, 12
+; RV32-NEXT:  .LBB15_14:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 9
-; RV32-NEXT:    vfmv.f.s ft1, v28
+; RV32-NEXT:    fsw ft0, 48(sp)
 ; RV32-NEXT:    bnez a0, .LBB15_16
 ; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.s ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 11
+; RV32-NEXT:    j .LBB15_17
 ; RV32-NEXT:  .LBB15_16:
-; RV32-NEXT:    fsw ft1, 36(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 8
+; RV32-NEXT:    vslidedown.vi v28, v8, 11
+; RV32-NEXT:  .LBB15_17:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 44(sp)
+; RV32-NEXT:    bnez a0, .LBB15_19
+; RV32-NEXT:  # %bb.18:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 10
+; RV32-NEXT:    j .LBB15_20
+; RV32-NEXT:  .LBB15_19:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v8, 10
+; RV32-NEXT:  .LBB15_20:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 40(sp)
+; RV32-NEXT:    bnez a0, .LBB15_22
+; RV32-NEXT:  # %bb.21:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 9
+; RV32-NEXT:    j .LBB15_23
+; RV32-NEXT:  .LBB15_22:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v8, 9
+; RV32-NEXT:  .LBB15_23:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 36(sp)
+; RV32-NEXT:    bnez a0, .LBB15_25
+; RV32-NEXT:  # %bb.24:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 8
+; RV32-NEXT:    j .LBB15_26
+; RV32-NEXT:  .LBB15_25:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 8
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_18
-; RV32-NEXT:  # %bb.17:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_18:
-; RV32-NEXT:    fsw ft1, 32(sp)
+; RV32-NEXT:  .LBB15_26:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 32(sp)
+; RV32-NEXT:    bnez a0, .LBB15_28
+; RV32-NEXT:  # %bb.27:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 7
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB15_29
+; RV32-NEXT:  .LBB15_28:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_20
-; RV32-NEXT:  # %bb.19:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_20:
-; RV32-NEXT:    fsw ft1, 28(sp)
+; RV32-NEXT:  .LBB15_29:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 28(sp)
+; RV32-NEXT:    bnez a0, .LBB15_31
+; RV32-NEXT:  # %bb.30:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 6
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB15_32
+; RV32-NEXT:  .LBB15_31:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_22
-; RV32-NEXT:  # %bb.21:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_22:
-; RV32-NEXT:    fsw ft1, 24(sp)
+; RV32-NEXT:  .LBB15_32:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB15_34
+; RV32-NEXT:  # %bb.33:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 5
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB15_35
+; RV32-NEXT:  .LBB15_34:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_24
-; RV32-NEXT:  # %bb.23:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_24:
-; RV32-NEXT:    fsw ft1, 20(sp)
+; RV32-NEXT:  .LBB15_35:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 20(sp)
+; RV32-NEXT:    bnez a0, .LBB15_37
+; RV32-NEXT:  # %bb.36:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 4
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB15_38
+; RV32-NEXT:  .LBB15_37:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_26
-; RV32-NEXT:  # %bb.25:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_26:
-; RV32-NEXT:    fsw ft1, 16(sp)
+; RV32-NEXT:  .LBB15_38:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB15_40
+; RV32-NEXT:  # %bb.39:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 3
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB15_41
+; RV32-NEXT:  .LBB15_40:
+; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_28
-; RV32-NEXT:  # %bb.27:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_28:
-; RV32-NEXT:    fsw ft1, 12(sp)
+; RV32-NEXT:  .LBB15_41:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 12(sp)
+; RV32-NEXT:    bnez a0, .LBB15_43
+; RV32-NEXT:  # %bb.42:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 2
-; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_30
-; RV32-NEXT:  # %bb.29:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_30:
-; RV32-NEXT:    fsw ft1, 8(sp)
+; RV32-NEXT:    j .LBB15_44
+; RV32-NEXT:  .LBB15_43:
 ; RV32-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    vslidedown.vi v28, v8, 2
+; RV32-NEXT:  .LBB15_44:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 8(sp)
+; RV32-NEXT:    bnez a0, .LBB15_46
+; RV32-NEXT:  # %bb.45:
+; RV32-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    j .LBB15_47
+; RV32-NEXT:  .LBB15_46:
+; RV32-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB15_32
-; RV32-NEXT:  # %bb.31:
-; RV32-NEXT:    fmv.s ft1, ft0
-; RV32-NEXT:  .LBB15_32:
-; RV32-NEXT:    fsw ft1, 4(sp)
+; RV32-NEXT:  .LBB15_47:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsw ft0, 4(sp)
 ; RV32-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
 ; RV32-NEXT:    vle32.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -128
@@ -2171,164 +2376,181 @@
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -64
 ; RV64-NEXT:    feq.s a0, fa0, fa1
-; RV64-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v12
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB15_2
+; RV64-NEXT:    bnez a0, .LBB15_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.s ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v12
+; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB15_4
 ; RV64-NEXT:  .LBB15_2:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v8, 15
+; RV64-NEXT:    j .LBB15_5
+; RV64-NEXT:  .LBB15_3:
+; RV64-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsw ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB15_2
+; RV64-NEXT:  .LBB15_4:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 15
+; RV64-NEXT:  .LBB15_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 15
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_4:
-; RV64-NEXT:    fsw ft1, 60(sp)
+; RV64-NEXT:    fsw ft0, 60(sp)
+; RV64-NEXT:    bnez a0, .LBB15_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 14
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_8
+; RV64-NEXT:  .LBB15_7:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 14
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_6:
-; RV64-NEXT:    fsw ft1, 56(sp)
+; RV64-NEXT:  .LBB15_8:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 56(sp)
+; RV64-NEXT:    bnez a0, .LBB15_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 13
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_11
+; RV64-NEXT:  .LBB15_10:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 13
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_8:
-; RV64-NEXT:    fsw ft1, 52(sp)
+; RV64-NEXT:  .LBB15_11:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 52(sp)
+; RV64-NEXT:    bnez a0, .LBB15_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 12
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_14
+; RV64-NEXT:  .LBB15_13:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 12
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_10:
-; RV64-NEXT:    fsw ft1, 48(sp)
+; RV64-NEXT:  .LBB15_14:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 48(sp)
+; RV64-NEXT:    bnez a0, .LBB15_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 11
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_17
+; RV64-NEXT:  .LBB15_16:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 11
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_12:
-; RV64-NEXT:    fsw ft1, 44(sp)
+; RV64-NEXT:  .LBB15_17:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 44(sp)
+; RV64-NEXT:    bnez a0, .LBB15_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 10
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_20
+; RV64-NEXT:  .LBB15_19:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 10
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_14:
-; RV64-NEXT:    fsw ft1, 40(sp)
+; RV64-NEXT:  .LBB15_20:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 40(sp)
+; RV64-NEXT:    bnez a0, .LBB15_22
+; RV64-NEXT:  # %bb.21:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 9
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_23
+; RV64-NEXT:  .LBB15_22:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 9
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_16:
-; RV64-NEXT:    fsw ft1, 36(sp)
+; RV64-NEXT:  .LBB15_23:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 36(sp)
+; RV64-NEXT:    bnez a0, .LBB15_25
+; RV64-NEXT:  # %bb.24:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 8
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_26
+; RV64-NEXT:  .LBB15_25:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 8
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_18
-; RV64-NEXT:  # %bb.17:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_18:
-; RV64-NEXT:    fsw ft1, 32(sp)
+; RV64-NEXT:  .LBB15_26:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 32(sp)
+; RV64-NEXT:    bnez a0, .LBB15_28
+; RV64-NEXT:  # %bb.27:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 7
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_29
+; RV64-NEXT:  .LBB15_28:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_20
-; RV64-NEXT:  # %bb.19:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_20:
-; RV64-NEXT:    fsw ft1, 28(sp)
+; RV64-NEXT:  .LBB15_29:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 28(sp)
+; RV64-NEXT:    bnez a0, .LBB15_31
+; RV64-NEXT:  # %bb.30:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 6
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_32
+; RV64-NEXT:  .LBB15_31:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_22
-; RV64-NEXT:  # %bb.21:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_22:
-; RV64-NEXT:    fsw ft1, 24(sp)
+; RV64-NEXT:  .LBB15_32:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB15_34
+; RV64-NEXT:  # %bb.33:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 5
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_35
+; RV64-NEXT:  .LBB15_34:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_24
-; RV64-NEXT:  # %bb.23:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_24:
-; RV64-NEXT:    fsw ft1, 20(sp)
+; RV64-NEXT:  .LBB15_35:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 20(sp)
+; RV64-NEXT:    bnez a0, .LBB15_37
+; RV64-NEXT:  # %bb.36:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 4
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_38
+; RV64-NEXT:  .LBB15_37:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_26
-; RV64-NEXT:  # %bb.25:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_26:
-; RV64-NEXT:    fsw ft1, 16(sp)
+; RV64-NEXT:  .LBB15_38:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB15_40
+; RV64-NEXT:  # %bb.39:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 3
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB15_41
+; RV64-NEXT:  .LBB15_40:
+; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_28
-; RV64-NEXT:  # %bb.27:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_28:
-; RV64-NEXT:    fsw ft1, 12(sp)
+; RV64-NEXT:  .LBB15_41:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 12(sp)
+; RV64-NEXT:    bnez a0, .LBB15_43
+; RV64-NEXT:  # %bb.42:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 2
-; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_30
-; RV64-NEXT:  # %bb.29:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_30:
-; RV64-NEXT:    fsw ft1, 8(sp)
+; RV64-NEXT:    j .LBB15_44
+; RV64-NEXT:  .LBB15_43:
 ; RV64-NEXT:    vsetivli a1, 1, e32,m4,ta,mu
-; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    vslidedown.vi v28, v8, 2
+; RV64-NEXT:  .LBB15_44:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 8(sp)
+; RV64-NEXT:    bnez a0, .LBB15_46
+; RV64-NEXT:  # %bb.45:
+; RV64-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    j .LBB15_47
+; RV64-NEXT:  .LBB15_46:
+; RV64-NEXT:    vsetivli a0, 1, e32,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB15_32
-; RV64-NEXT:  # %bb.31:
-; RV64-NEXT:    fmv.s ft1, ft0
-; RV64-NEXT:  .LBB15_32:
-; RV64-NEXT:    fsw ft1, 4(sp)
+; RV64-NEXT:  .LBB15_47:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsw ft0, 4(sp)
 ; RV64-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
 ; RV64-NEXT:    vle32.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -128
@@ -2344,20 +2566,22 @@
 define <2 x double> @select_v2f64(i1 zeroext %c, <2 x double> %a, <2 x double> %b) {
 ; CHECK-LABEL: select_v2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
-; CHECK-NEXT:    vfmv.f.s ft1, v9
-; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    vslidedown.vi v25, v9, 1
-; CHECK-NEXT:    vfmv.f.s ft3, v25
-; CHECK-NEXT:    vslidedown.vi v25, v8, 1
-; CHECK-NEXT:    vfmv.f.s ft2, v25
 ; CHECK-NEXT:    bnez a0, .LBB16_2
 ; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.d ft0, ft1
-; CHECK-NEXT:    fmv.d ft2, ft3
+; CHECK-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    vsetivli a0, 1, e64,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v9, 1
+; CHECK-NEXT:    j .LBB16_3
 ; CHECK-NEXT:  .LBB16_2:
+; CHECK-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v8
+; CHECK-NEXT:    vsetivli a0, 1, e64,m1,ta,mu
+; CHECK-NEXT:    vslidedown.vi v25, v8, 1
+; CHECK-NEXT:  .LBB16_3:
+; CHECK-NEXT:    vfmv.f.s ft1, v25
 ; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
-; CHECK-NEXT:    vfmv.v.f v8, ft2
+; CHECK-NEXT:    vfmv.v.f v8, ft1
 ; CHECK-NEXT:    vfmv.s.f v8, ft0
 ; CHECK-NEXT:    ret
   %v = select i1 %c, <2 x double> %a, <2 x double> %b
@@ -2368,23 +2592,27 @@
 ; CHECK-LABEL: selectcc_v2f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    feq.d a0, fa0, fa1
+; CHECK-NEXT:    bnez a0, .LBB17_2
+; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v9, 1
-; CHECK-NEXT:    vfmv.f.s ft1, v25
+; CHECK-NEXT:    j .LBB17_3
+; CHECK-NEXT:  .LBB17_2:
+; CHECK-NEXT:    vsetivli a1, 1, e64,m1,ta,mu
 ; CHECK-NEXT:    vslidedown.vi v25, v8, 1
+; CHECK-NEXT:  .LBB17_3:
 ; CHECK-NEXT:    vfmv.f.s ft0, v25
-; CHECK-NEXT:    bnez a0, .LBB17_2
-; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    fmv.d ft0, ft1
-; CHECK-NEXT:  .LBB17_2:
 ; CHECK-NEXT:    vsetivli a1, 2, e64,m1,ta,mu
 ; CHECK-NEXT:    vfmv.v.f v25, ft0
-; CHECK-NEXT:    vfmv.f.s ft1, v9
+; CHECK-NEXT:    bnez a0, .LBB17_5
+; CHECK-NEXT:  # %bb.4:
+; CHECK-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vfmv.f.s ft0, v9
+; CHECK-NEXT:    j .LBB17_6
+; CHECK-NEXT:  .LBB17_5:
+; CHECK-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
-; CHECK-NEXT:    bnez a0, .LBB17_4
-; CHECK-NEXT:  # %bb.3:
-; CHECK-NEXT:    fmv.d ft0, ft1
-; CHECK-NEXT:  .LBB17_4:
+; CHECK-NEXT:  .LBB17_6:
 ; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
 ; CHECK-NEXT:    vfmv.s.f v25, ft0
 ; CHECK-NEXT:    vmv1r.v v8, v25
@@ -2406,44 +2634,49 @@
 ; RV32-NEXT:    addi s0, sp, 64
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -32
-; RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v10
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB18_2
+; RV32-NEXT:    bnez a0, .LBB18_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.d ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v10
+; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB18_4
 ; RV32-NEXT:  .LBB18_2:
+; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 3
+; RV32-NEXT:    j .LBB18_5
+; RV32-NEXT:  .LBB18_3:
+; RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB18_2
+; RV32-NEXT:  .LBB18_4:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 3
+; RV32-NEXT:  .LBB18_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB18_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB18_4:
-; RV32-NEXT:    fsd ft1, 24(sp)
+; RV32-NEXT:    fsd ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB18_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 2
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB18_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB18_6:
-; RV32-NEXT:    fsd ft1, 16(sp)
+; RV32-NEXT:    j .LBB18_8
+; RV32-NEXT:  .LBB18_7:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    vslidedown.vi v26, v8, 2
+; RV32-NEXT:  .LBB18_8:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsd ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB18_10
+; RV32-NEXT:  # %bb.9:
+; RV32-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    j .LBB18_11
+; RV32-NEXT:  .LBB18_10:
+; RV32-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB18_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB18_8:
-; RV32-NEXT:    fsd ft1, 8(sp)
+; RV32-NEXT:  .LBB18_11:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsd ft0, 8(sp)
 ; RV32-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
 ; RV32-NEXT:    vle64.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -64
@@ -2463,44 +2696,49 @@
 ; RV64-NEXT:    addi s0, sp, 64
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -32
-; RV64-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v10
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB18_2
+; RV64-NEXT:    bnez a0, .LBB18_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.d ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v10
+; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB18_4
 ; RV64-NEXT:  .LBB18_2:
+; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 3
+; RV64-NEXT:    j .LBB18_5
+; RV64-NEXT:  .LBB18_3:
+; RV64-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB18_2
+; RV64-NEXT:  .LBB18_4:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 3
+; RV64-NEXT:  .LBB18_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB18_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB18_4:
-; RV64-NEXT:    fsd ft1, 24(sp)
+; RV64-NEXT:    fsd ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB18_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 2
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB18_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB18_6:
-; RV64-NEXT:    fsd ft1, 16(sp)
+; RV64-NEXT:    j .LBB18_8
+; RV64-NEXT:  .LBB18_7:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    vslidedown.vi v26, v8, 2
+; RV64-NEXT:  .LBB18_8:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsd ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB18_10
+; RV64-NEXT:  # %bb.9:
+; RV64-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    j .LBB18_11
+; RV64-NEXT:  .LBB18_10:
+; RV64-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB18_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB18_8:
-; RV64-NEXT:    fsd ft1, 8(sp)
+; RV64-NEXT:  .LBB18_11:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsd ft0, 8(sp)
 ; RV64-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
 ; RV64-NEXT:    vle64.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -64
@@ -2525,44 +2763,49 @@
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -32
 ; RV32-NEXT:    feq.d a0, fa0, fa1
-; RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v10
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB19_2
+; RV32-NEXT:    bnez a0, .LBB19_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.d ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v10
+; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB19_4
 ; RV32-NEXT:  .LBB19_2:
+; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v8, 3
+; RV32-NEXT:    j .LBB19_5
+; RV32-NEXT:  .LBB19_3:
+; RV32-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB19_2
+; RV32-NEXT:  .LBB19_4:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 3
+; RV32-NEXT:  .LBB19_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB19_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB19_4:
-; RV32-NEXT:    fsd ft1, 24(sp)
+; RV32-NEXT:    fsd ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB19_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v10, 2
-; RV32-NEXT:    vfmv.f.s ft0, v26
-; RV32-NEXT:    vslidedown.vi v26, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB19_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB19_6:
-; RV32-NEXT:    fsd ft1, 16(sp)
+; RV32-NEXT:    j .LBB19_8
+; RV32-NEXT:  .LBB19_7:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
-; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    vslidedown.vi v26, v8, 2
+; RV32-NEXT:  .LBB19_8:
 ; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsd ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB19_10
+; RV32-NEXT:  # %bb.9:
+; RV32-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
+; RV32-NEXT:    vslidedown.vi v26, v10, 1
+; RV32-NEXT:    j .LBB19_11
+; RV32-NEXT:  .LBB19_10:
+; RV32-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
 ; RV32-NEXT:    vslidedown.vi v26, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v26
-; RV32-NEXT:    bnez a0, .LBB19_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB19_8:
-; RV32-NEXT:    fsd ft1, 8(sp)
+; RV32-NEXT:  .LBB19_11:
+; RV32-NEXT:    vfmv.f.s ft0, v26
+; RV32-NEXT:    fsd ft0, 8(sp)
 ; RV32-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
 ; RV32-NEXT:    vle64.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -64
@@ -2583,44 +2826,49 @@
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -32
 ; RV64-NEXT:    feq.d a0, fa0, fa1
-; RV64-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v10
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB19_2
+; RV64-NEXT:    bnez a0, .LBB19_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.d ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v10
+; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB19_4
 ; RV64-NEXT:  .LBB19_2:
+; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v8, 3
+; RV64-NEXT:    j .LBB19_5
+; RV64-NEXT:  .LBB19_3:
+; RV64-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB19_2
+; RV64-NEXT:  .LBB19_4:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 3
+; RV64-NEXT:  .LBB19_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB19_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB19_4:
-; RV64-NEXT:    fsd ft1, 24(sp)
+; RV64-NEXT:    fsd ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB19_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v10, 2
-; RV64-NEXT:    vfmv.f.s ft0, v26
-; RV64-NEXT:    vslidedown.vi v26, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB19_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB19_6:
-; RV64-NEXT:    fsd ft1, 16(sp)
+; RV64-NEXT:    j .LBB19_8
+; RV64-NEXT:  .LBB19_7:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m2,ta,mu
-; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    vslidedown.vi v26, v8, 2
+; RV64-NEXT:  .LBB19_8:
 ; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsd ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB19_10
+; RV64-NEXT:  # %bb.9:
+; RV64-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
+; RV64-NEXT:    vslidedown.vi v26, v10, 1
+; RV64-NEXT:    j .LBB19_11
+; RV64-NEXT:  .LBB19_10:
+; RV64-NEXT:    vsetivli a0, 1, e64,m2,ta,mu
 ; RV64-NEXT:    vslidedown.vi v26, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v26
-; RV64-NEXT:    bnez a0, .LBB19_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB19_8:
-; RV64-NEXT:    fsd ft1, 8(sp)
+; RV64-NEXT:  .LBB19_11:
+; RV64-NEXT:    vfmv.f.s ft0, v26
+; RV64-NEXT:    fsd ft0, 8(sp)
 ; RV64-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
 ; RV64-NEXT:    vle64.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -64
@@ -2645,84 +2893,93 @@
 ; RV32-NEXT:    addi s0, sp, 128
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -64
-; RV32-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v12
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB20_2
+; RV32-NEXT:    bnez a0, .LBB20_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.d ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v12
+; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB20_4
 ; RV32-NEXT:  .LBB20_2:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v8, 7
+; RV32-NEXT:    j .LBB20_5
+; RV32-NEXT:  .LBB20_3:
+; RV32-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB20_2
+; RV32-NEXT:  .LBB20_4:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 7
+; RV32-NEXT:  .LBB20_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB20_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB20_4:
-; RV32-NEXT:    fsd ft1, 56(sp)
+; RV32-NEXT:    fsd ft0, 56(sp)
+; RV32-NEXT:    bnez a0, .LBB20_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 6
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB20_8
+; RV32-NEXT:  .LBB20_7:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB20_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB20_6:
-; RV32-NEXT:    fsd ft1, 48(sp)
+; RV32-NEXT:  .LBB20_8:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 48(sp)
+; RV32-NEXT:    bnez a0, .LBB20_10
+; RV32-NEXT:  # %bb.9:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 5
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB20_11
+; RV32-NEXT:  .LBB20_10:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB20_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB20_8:
-; RV32-NEXT:    fsd ft1, 40(sp)
+; RV32-NEXT:  .LBB20_11:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 40(sp)
+; RV32-NEXT:    bnez a0, .LBB20_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 4
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB20_14
+; RV32-NEXT:  .LBB20_13:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB20_10
-; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB20_10:
-; RV32-NEXT:    fsd ft1, 32(sp)
+; RV32-NEXT:  .LBB20_14:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 32(sp)
+; RV32-NEXT:    bnez a0, .LBB20_16
+; RV32-NEXT:  # %bb.15:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 3
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB20_17
+; RV32-NEXT:  .LBB20_16:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB20_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB20_12:
-; RV32-NEXT:    fsd ft1, 24(sp)
+; RV32-NEXT:  .LBB20_17:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB20_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 2
-; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB20_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB20_14:
-; RV32-NEXT:    fsd ft1, 16(sp)
+; RV32-NEXT:    j .LBB20_20
+; RV32-NEXT:  .LBB20_19:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    vslidedown.vi v28, v8, 2
+; RV32-NEXT:  .LBB20_20:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB20_22
+; RV32-NEXT:  # %bb.21:
+; RV32-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    j .LBB20_23
+; RV32-NEXT:  .LBB20_22:
+; RV32-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB20_16
-; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB20_16:
-; RV32-NEXT:    fsd ft1, 8(sp)
+; RV32-NEXT:  .LBB20_23:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 8(sp)
 ; RV32-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
 ; RV32-NEXT:    vle64.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -128
@@ -2742,84 +2999,93 @@
 ; RV64-NEXT:    addi s0, sp, 128
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -64
-; RV64-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v12
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB20_2
+; RV64-NEXT:    bnez a0, .LBB20_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.d ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v12
+; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB20_4
 ; RV64-NEXT:  .LBB20_2:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v8, 7
+; RV64-NEXT:    j .LBB20_5
+; RV64-NEXT:  .LBB20_3:
+; RV64-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB20_2
+; RV64-NEXT:  .LBB20_4:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 7
+; RV64-NEXT:  .LBB20_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB20_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB20_4:
-; RV64-NEXT:    fsd ft1, 56(sp)
+; RV64-NEXT:    fsd ft0, 56(sp)
+; RV64-NEXT:    bnez a0, .LBB20_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 6
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB20_8
+; RV64-NEXT:  .LBB20_7:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB20_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB20_6:
-; RV64-NEXT:    fsd ft1, 48(sp)
+; RV64-NEXT:  .LBB20_8:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 48(sp)
+; RV64-NEXT:    bnez a0, .LBB20_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 5
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB20_11
+; RV64-NEXT:  .LBB20_10:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB20_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB20_8:
-; RV64-NEXT:    fsd ft1, 40(sp)
+; RV64-NEXT:  .LBB20_11:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 40(sp)
+; RV64-NEXT:    bnez a0, .LBB20_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 4
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB20_14
+; RV64-NEXT:  .LBB20_13:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB20_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB20_10:
-; RV64-NEXT:    fsd ft1, 32(sp)
+; RV64-NEXT:  .LBB20_14:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 32(sp)
+; RV64-NEXT:    bnez a0, .LBB20_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 3
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB20_17
+; RV64-NEXT:  .LBB20_16:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB20_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB20_12:
-; RV64-NEXT:    fsd ft1, 24(sp)
+; RV64-NEXT:  .LBB20_17:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB20_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 2
-; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB20_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB20_14:
-; RV64-NEXT:    fsd ft1, 16(sp)
+; RV64-NEXT:    j .LBB20_20
+; RV64-NEXT:  .LBB20_19:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
-; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    vslidedown.vi v28, v8, 2
+; RV64-NEXT:  .LBB20_20:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB20_22
+; RV64-NEXT:  # %bb.21:
+; RV64-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    j .LBB20_23
+; RV64-NEXT:  .LBB20_22:
+; RV64-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB20_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB20_16:
-; RV64-NEXT:    fsd ft1, 8(sp)
+; RV64-NEXT:  .LBB20_23:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 8(sp)
 ; RV64-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
 ; RV64-NEXT:    vle64.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -128
@@ -2844,84 +3110,93 @@
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -64
 ; RV32-NEXT:    feq.d a0, fa0, fa1
-; RV32-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v12
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB21_2
+; RV32-NEXT:    bnez a0, .LBB21_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.d ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v12
+; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB21_4
 ; RV32-NEXT:  .LBB21_2:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v8, 7
+; RV32-NEXT:    j .LBB21_5
+; RV32-NEXT:  .LBB21_3:
+; RV32-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB21_2
+; RV32-NEXT:  .LBB21_4:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 7
+; RV32-NEXT:  .LBB21_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB21_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB21_4:
-; RV32-NEXT:    fsd ft1, 56(sp)
+; RV32-NEXT:    fsd ft0, 56(sp)
+; RV32-NEXT:    bnez a0, .LBB21_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 6
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB21_8
+; RV32-NEXT:  .LBB21_7:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB21_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB21_6:
-; RV32-NEXT:    fsd ft1, 48(sp)
+; RV32-NEXT:  .LBB21_8:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 48(sp)
+; RV32-NEXT:    bnez a0, .LBB21_10
+; RV32-NEXT:  # %bb.9:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 5
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB21_11
+; RV32-NEXT:  .LBB21_10:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB21_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB21_8:
-; RV32-NEXT:    fsd ft1, 40(sp)
+; RV32-NEXT:  .LBB21_11:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 40(sp)
+; RV32-NEXT:    bnez a0, .LBB21_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 4
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB21_14
+; RV32-NEXT:  .LBB21_13:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB21_10
-; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB21_10:
-; RV32-NEXT:    fsd ft1, 32(sp)
+; RV32-NEXT:  .LBB21_14:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 32(sp)
+; RV32-NEXT:    bnez a0, .LBB21_16
+; RV32-NEXT:  # %bb.15:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 3
-; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    j .LBB21_17
+; RV32-NEXT:  .LBB21_16:
+; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB21_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB21_12:
-; RV32-NEXT:    fsd ft1, 24(sp)
+; RV32-NEXT:  .LBB21_17:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB21_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v12, 2
-; RV32-NEXT:    vfmv.f.s ft0, v28
-; RV32-NEXT:    vslidedown.vi v28, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB21_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB21_14:
-; RV32-NEXT:    fsd ft1, 16(sp)
+; RV32-NEXT:    j .LBB21_20
+; RV32-NEXT:  .LBB21_19:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
-; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    vslidedown.vi v28, v8, 2
+; RV32-NEXT:  .LBB21_20:
 ; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB21_22
+; RV32-NEXT:  # %bb.21:
+; RV32-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
+; RV32-NEXT:    vslidedown.vi v28, v12, 1
+; RV32-NEXT:    j .LBB21_23
+; RV32-NEXT:  .LBB21_22:
+; RV32-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
 ; RV32-NEXT:    vslidedown.vi v28, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v28
-; RV32-NEXT:    bnez a0, .LBB21_16
-; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB21_16:
-; RV32-NEXT:    fsd ft1, 8(sp)
+; RV32-NEXT:  .LBB21_23:
+; RV32-NEXT:    vfmv.f.s ft0, v28
+; RV32-NEXT:    fsd ft0, 8(sp)
 ; RV32-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
 ; RV32-NEXT:    vle64.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -128
@@ -2942,84 +3217,93 @@
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -64
 ; RV64-NEXT:    feq.d a0, fa0, fa1
-; RV64-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v12
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB21_2
+; RV64-NEXT:    bnez a0, .LBB21_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.d ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v12
+; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB21_4
 ; RV64-NEXT:  .LBB21_2:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v8, 7
+; RV64-NEXT:    j .LBB21_5
+; RV64-NEXT:  .LBB21_3:
+; RV64-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB21_2
+; RV64-NEXT:  .LBB21_4:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 7
+; RV64-NEXT:  .LBB21_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB21_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB21_4:
-; RV64-NEXT:    fsd ft1, 56(sp)
+; RV64-NEXT:    fsd ft0, 56(sp)
+; RV64-NEXT:    bnez a0, .LBB21_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 6
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB21_8
+; RV64-NEXT:  .LBB21_7:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB21_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB21_6:
-; RV64-NEXT:    fsd ft1, 48(sp)
+; RV64-NEXT:  .LBB21_8:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 48(sp)
+; RV64-NEXT:    bnez a0, .LBB21_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 5
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB21_11
+; RV64-NEXT:  .LBB21_10:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB21_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB21_8:
-; RV64-NEXT:    fsd ft1, 40(sp)
+; RV64-NEXT:  .LBB21_11:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 40(sp)
+; RV64-NEXT:    bnez a0, .LBB21_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 4
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB21_14
+; RV64-NEXT:  .LBB21_13:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB21_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB21_10:
-; RV64-NEXT:    fsd ft1, 32(sp)
+; RV64-NEXT:  .LBB21_14:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 32(sp)
+; RV64-NEXT:    bnez a0, .LBB21_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 3
-; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    j .LBB21_17
+; RV64-NEXT:  .LBB21_16:
+; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB21_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB21_12:
-; RV64-NEXT:    fsd ft1, 24(sp)
+; RV64-NEXT:  .LBB21_17:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB21_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v12, 2
-; RV64-NEXT:    vfmv.f.s ft0, v28
-; RV64-NEXT:    vslidedown.vi v28, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB21_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB21_14:
-; RV64-NEXT:    fsd ft1, 16(sp)
+; RV64-NEXT:    j .LBB21_20
+; RV64-NEXT:  .LBB21_19:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m4,ta,mu
-; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    vslidedown.vi v28, v8, 2
+; RV64-NEXT:  .LBB21_20:
 ; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB21_22
+; RV64-NEXT:  # %bb.21:
+; RV64-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
+; RV64-NEXT:    vslidedown.vi v28, v12, 1
+; RV64-NEXT:    j .LBB21_23
+; RV64-NEXT:  .LBB21_22:
+; RV64-NEXT:    vsetivli a0, 1, e64,m4,ta,mu
 ; RV64-NEXT:    vslidedown.vi v28, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v28
-; RV64-NEXT:    bnez a0, .LBB21_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB21_16:
-; RV64-NEXT:    fsd ft1, 8(sp)
+; RV64-NEXT:  .LBB21_23:
+; RV64-NEXT:    vfmv.f.s ft0, v28
+; RV64-NEXT:    fsd ft0, 8(sp)
 ; RV64-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
 ; RV64-NEXT:    vle64.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -128
@@ -3044,164 +3328,181 @@
 ; RV32-NEXT:    addi s0, sp, 256
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -128
-; RV32-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v16
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB22_2
+; RV32-NEXT:    bnez a0, .LBB22_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.d ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v16
+; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB22_4
 ; RV32-NEXT:  .LBB22_2:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v8, 15
+; RV32-NEXT:    j .LBB22_5
+; RV32-NEXT:  .LBB22_3:
+; RV32-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB22_2
+; RV32-NEXT:  .LBB22_4:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 15
+; RV32-NEXT:  .LBB22_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 15
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_4:
-; RV32-NEXT:    fsd ft1, 120(sp)
+; RV32-NEXT:    fsd ft0, 120(sp)
+; RV32-NEXT:    bnez a0, .LBB22_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 14
-; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 14
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_6:
-; RV32-NEXT:    fsd ft1, 112(sp)
+; RV32-NEXT:    j .LBB22_8
+; RV32-NEXT:  .LBB22_7:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 13
-; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 13
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.d ft1, ft0
+; RV32-NEXT:    vslidedown.vi v24, v8, 14
 ; RV32-NEXT:  .LBB22_8:
-; RV32-NEXT:    fsd ft1, 104(sp)
-; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 12
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 12
-; RV32-NEXT:    vfmv.f.s ft1, v24
+; RV32-NEXT:    fsd ft0, 112(sp)
 ; RV32-NEXT:    bnez a0, .LBB22_10
 ; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.d ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 13
+; RV32-NEXT:    j .LBB22_11
 ; RV32-NEXT:  .LBB22_10:
-; RV32-NEXT:    fsd ft1, 96(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 11
+; RV32-NEXT:    vslidedown.vi v24, v8, 13
+; RV32-NEXT:  .LBB22_11:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 11
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_12:
-; RV32-NEXT:    fsd ft1, 88(sp)
+; RV32-NEXT:    fsd ft0, 104(sp)
+; RV32-NEXT:    bnez a0, .LBB22_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 10
-; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 10
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_14:
-; RV32-NEXT:    fsd ft1, 80(sp)
+; RV32-NEXT:    vslidedown.vi v24, v16, 12
+; RV32-NEXT:    j .LBB22_14
+; RV32-NEXT:  .LBB22_13:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 9
+; RV32-NEXT:    vslidedown.vi v24, v8, 12
+; RV32-NEXT:  .LBB22_14:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 9
-; RV32-NEXT:    vfmv.f.s ft1, v24
+; RV32-NEXT:    fsd ft0, 96(sp)
 ; RV32-NEXT:    bnez a0, .LBB22_16
 ; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.d ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 11
+; RV32-NEXT:    j .LBB22_17
 ; RV32-NEXT:  .LBB22_16:
-; RV32-NEXT:    fsd ft1, 72(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 8
+; RV32-NEXT:    vslidedown.vi v24, v8, 11
+; RV32-NEXT:  .LBB22_17:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 8
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_18
-; RV32-NEXT:  # %bb.17:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_18:
-; RV32-NEXT:    fsd ft1, 64(sp)
+; RV32-NEXT:    fsd ft0, 88(sp)
+; RV32-NEXT:    bnez a0, .LBB22_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 7
-; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_20
-; RV32-NEXT:  # %bb.19:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_20:
-; RV32-NEXT:    fsd ft1, 56(sp)
+; RV32-NEXT:    vslidedown.vi v24, v16, 10
+; RV32-NEXT:    j .LBB22_20
+; RV32-NEXT:  .LBB22_19:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 6
+; RV32-NEXT:    vslidedown.vi v24, v8, 10
+; RV32-NEXT:  .LBB22_20:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v24
+; RV32-NEXT:    fsd ft0, 80(sp)
 ; RV32-NEXT:    bnez a0, .LBB22_22
 ; RV32-NEXT:  # %bb.21:
-; RV32-NEXT:    fmv.d ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 9
+; RV32-NEXT:    j .LBB22_23
 ; RV32-NEXT:  .LBB22_22:
-; RV32-NEXT:    fsd ft1, 48(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 5
+; RV32-NEXT:    vslidedown.vi v24, v8, 9
+; RV32-NEXT:  .LBB22_23:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_24
-; RV32-NEXT:  # %bb.23:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_24:
-; RV32-NEXT:    fsd ft1, 40(sp)
+; RV32-NEXT:    fsd ft0, 72(sp)
+; RV32-NEXT:    bnez a0, .LBB22_25
+; RV32-NEXT:  # %bb.24:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 4
-; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_26
-; RV32-NEXT:  # %bb.25:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_26:
-; RV32-NEXT:    fsd ft1, 32(sp)
+; RV32-NEXT:    vslidedown.vi v24, v16, 8
+; RV32-NEXT:    j .LBB22_26
+; RV32-NEXT:  .LBB22_25:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 3
+; RV32-NEXT:    vslidedown.vi v24, v8, 8
+; RV32-NEXT:  .LBB22_26:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v24
+; RV32-NEXT:    fsd ft0, 64(sp)
 ; RV32-NEXT:    bnez a0, .LBB22_28
 ; RV32-NEXT:  # %bb.27:
-; RV32-NEXT:    fmv.d ft1, ft0
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 7
+; RV32-NEXT:    j .LBB22_29
 ; RV32-NEXT:  .LBB22_28:
-; RV32-NEXT:    fsd ft1, 24(sp)
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v24, v16, 2
+; RV32-NEXT:    vslidedown.vi v24, v8, 7
+; RV32-NEXT:  .LBB22_29:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB22_30
-; RV32-NEXT:  # %bb.29:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB22_30:
-; RV32-NEXT:    fsd ft1, 16(sp)
+; RV32-NEXT:    fsd ft0, 56(sp)
+; RV32-NEXT:    bnez a0, .LBB22_31
+; RV32-NEXT:  # %bb.30:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v16, v16, 1
-; RV32-NEXT:    vfmv.f.s ft0, v16
-; RV32-NEXT:    vslidedown.vi v8, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v8
-; RV32-NEXT:    bnez a0, .LBB22_32
-; RV32-NEXT:  # %bb.31:
-; RV32-NEXT:    fmv.d ft1, ft0
+; RV32-NEXT:    vslidedown.vi v24, v16, 6
+; RV32-NEXT:    j .LBB22_32
+; RV32-NEXT:  .LBB22_31:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v8, 6
 ; RV32-NEXT:  .LBB22_32:
-; RV32-NEXT:    fsd ft1, 8(sp)
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 48(sp)
+; RV32-NEXT:    bnez a0, .LBB22_34
+; RV32-NEXT:  # %bb.33:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 5
+; RV32-NEXT:    j .LBB22_35
+; RV32-NEXT:  .LBB22_34:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v8, 5
+; RV32-NEXT:  .LBB22_35:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 40(sp)
+; RV32-NEXT:    bnez a0, .LBB22_37
+; RV32-NEXT:  # %bb.36:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 4
+; RV32-NEXT:    j .LBB22_38
+; RV32-NEXT:  .LBB22_37:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v8, 4
+; RV32-NEXT:  .LBB22_38:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 32(sp)
+; RV32-NEXT:    bnez a0, .LBB22_40
+; RV32-NEXT:  # %bb.39:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 3
+; RV32-NEXT:    j .LBB22_41
+; RV32-NEXT:  .LBB22_40:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v8, 3
+; RV32-NEXT:  .LBB22_41:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB22_43
+; RV32-NEXT:  # %bb.42:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v16, 2
+; RV32-NEXT:    j .LBB22_44
+; RV32-NEXT:  .LBB22_43:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v8, 2
+; RV32-NEXT:  .LBB22_44:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB22_46
+; RV32-NEXT:  # %bb.45:
+; RV32-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v8, v16, 1
+; RV32-NEXT:    j .LBB22_47
+; RV32-NEXT:  .LBB22_46:
+; RV32-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:  .LBB22_47:
+; RV32-NEXT:    vfmv.f.s ft0, v8
+; RV32-NEXT:    fsd ft0, 8(sp)
 ; RV32-NEXT:    vsetivli a0, 16, e64,m8,ta,mu
 ; RV32-NEXT:    vle64.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -256
@@ -3221,164 +3522,181 @@
 ; RV64-NEXT:    addi s0, sp, 256
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -128
-; RV64-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v16
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB22_2
+; RV64-NEXT:    bnez a0, .LBB22_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.d ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v16
+; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB22_4
 ; RV64-NEXT:  .LBB22_2:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV64-NEXT:    vslidedown.vi v24, v8, 15
+; RV64-NEXT:    j .LBB22_5
+; RV64-NEXT:  .LBB22_3:
+; RV64-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB22_2
+; RV64-NEXT:  .LBB22_4:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 15
+; RV64-NEXT:  .LBB22_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v24
-; RV64-NEXT:    vslidedown.vi v24, v8, 15
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_4:
-; RV64-NEXT:    fsd ft1, 120(sp)
+; RV64-NEXT:    fsd ft0, 120(sp)
+; RV64-NEXT:    bnez a0, .LBB22_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 14
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_8
+; RV64-NEXT:  .LBB22_7:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 14
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_6:
-; RV64-NEXT:    fsd ft1, 112(sp)
+; RV64-NEXT:  .LBB22_8:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 112(sp)
+; RV64-NEXT:    bnez a0, .LBB22_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 13
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_11
+; RV64-NEXT:  .LBB22_10:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 13
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_8:
-; RV64-NEXT:    fsd ft1, 104(sp)
+; RV64-NEXT:  .LBB22_11:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 104(sp)
+; RV64-NEXT:    bnez a0, .LBB22_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 12
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_14
+; RV64-NEXT:  .LBB22_13:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 12
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_10:
-; RV64-NEXT:    fsd ft1, 96(sp)
+; RV64-NEXT:  .LBB22_14:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 96(sp)
+; RV64-NEXT:    bnez a0, .LBB22_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 11
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_17
+; RV64-NEXT:  .LBB22_16:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 11
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_12:
-; RV64-NEXT:    fsd ft1, 88(sp)
+; RV64-NEXT:  .LBB22_17:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 88(sp)
+; RV64-NEXT:    bnez a0, .LBB22_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 10
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_20
+; RV64-NEXT:  .LBB22_19:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 10
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_14:
-; RV64-NEXT:    fsd ft1, 80(sp)
+; RV64-NEXT:  .LBB22_20:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 80(sp)
+; RV64-NEXT:    bnez a0, .LBB22_22
+; RV64-NEXT:  # %bb.21:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 9
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_23
+; RV64-NEXT:  .LBB22_22:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 9
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_16:
-; RV64-NEXT:    fsd ft1, 72(sp)
+; RV64-NEXT:  .LBB22_23:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 72(sp)
+; RV64-NEXT:    bnez a0, .LBB22_25
+; RV64-NEXT:  # %bb.24:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 8
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_26
+; RV64-NEXT:  .LBB22_25:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 8
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_18
-; RV64-NEXT:  # %bb.17:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_18:
-; RV64-NEXT:    fsd ft1, 64(sp)
+; RV64-NEXT:  .LBB22_26:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 64(sp)
+; RV64-NEXT:    bnez a0, .LBB22_28
+; RV64-NEXT:  # %bb.27:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 7
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_29
+; RV64-NEXT:  .LBB22_28:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_20
-; RV64-NEXT:  # %bb.19:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_20:
-; RV64-NEXT:    fsd ft1, 56(sp)
+; RV64-NEXT:  .LBB22_29:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 56(sp)
+; RV64-NEXT:    bnez a0, .LBB22_31
+; RV64-NEXT:  # %bb.30:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 6
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_32
+; RV64-NEXT:  .LBB22_31:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_22
-; RV64-NEXT:  # %bb.21:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_22:
-; RV64-NEXT:    fsd ft1, 48(sp)
+; RV64-NEXT:  .LBB22_32:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 48(sp)
+; RV64-NEXT:    bnez a0, .LBB22_34
+; RV64-NEXT:  # %bb.33:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 5
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_35
+; RV64-NEXT:  .LBB22_34:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_24
-; RV64-NEXT:  # %bb.23:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_24:
-; RV64-NEXT:    fsd ft1, 40(sp)
+; RV64-NEXT:  .LBB22_35:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 40(sp)
+; RV64-NEXT:    bnez a0, .LBB22_37
+; RV64-NEXT:  # %bb.36:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 4
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_38
+; RV64-NEXT:  .LBB22_37:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_26
-; RV64-NEXT:  # %bb.25:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_26:
-; RV64-NEXT:    fsd ft1, 32(sp)
+; RV64-NEXT:  .LBB22_38:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 32(sp)
+; RV64-NEXT:    bnez a0, .LBB22_40
+; RV64-NEXT:  # %bb.39:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 3
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB22_41
+; RV64-NEXT:  .LBB22_40:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_28
-; RV64-NEXT:  # %bb.27:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_28:
-; RV64-NEXT:    fsd ft1, 24(sp)
+; RV64-NEXT:  .LBB22_41:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB22_43
+; RV64-NEXT:  # %bb.42:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 2
-; RV64-NEXT:    vfmv.f.s ft0, v24
-; RV64-NEXT:    vslidedown.vi v24, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB22_30
-; RV64-NEXT:  # %bb.29:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_30:
-; RV64-NEXT:    fsd ft1, 16(sp)
+; RV64-NEXT:    j .LBB22_44
+; RV64-NEXT:  .LBB22_43:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV64-NEXT:    vslidedown.vi v16, v16, 1
-; RV64-NEXT:    vfmv.f.s ft0, v16
+; RV64-NEXT:    vslidedown.vi v24, v8, 2
+; RV64-NEXT:  .LBB22_44:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB22_46
+; RV64-NEXT:  # %bb.45:
+; RV64-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
+; RV64-NEXT:    vslidedown.vi v8, v16, 1
+; RV64-NEXT:    j .LBB22_47
+; RV64-NEXT:  .LBB22_46:
+; RV64-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v8, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v8
-; RV64-NEXT:    bnez a0, .LBB22_32
-; RV64-NEXT:  # %bb.31:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB22_32:
-; RV64-NEXT:    fsd ft1, 8(sp)
+; RV64-NEXT:  .LBB22_47:
+; RV64-NEXT:    vfmv.f.s ft0, v8
+; RV64-NEXT:    fsd ft0, 8(sp)
 ; RV64-NEXT:    vsetivli a0, 16, e64,m8,ta,mu
 ; RV64-NEXT:    vle64.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -256
@@ -3403,164 +3721,181 @@
 ; RV32-NEXT:    .cfi_def_cfa s0, 0
 ; RV32-NEXT:    andi sp, sp, -128
 ; RV32-NEXT:    feq.d a0, fa0, fa1
-; RV32-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
-; RV32-NEXT:    vfmv.f.s ft1, v16
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    bnez a0, .LBB23_2
+; RV32-NEXT:    bnez a0, .LBB23_3
 ; RV32-NEXT:  # %bb.1:
-; RV32-NEXT:    fmv.d ft0, ft1
+; RV32-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v16
+; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    beqz a0, .LBB23_4
 ; RV32-NEXT:  .LBB23_2:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v24, v8, 15
+; RV32-NEXT:    j .LBB23_5
+; RV32-NEXT:  .LBB23_3:
+; RV32-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV32-NEXT:    vfmv.f.s ft0, v8
 ; RV32-NEXT:    fsd ft0, 0(sp)
+; RV32-NEXT:    bnez a0, .LBB23_2
+; RV32-NEXT:  .LBB23_4:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 15
+; RV32-NEXT:  .LBB23_5:
 ; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 15
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_4
-; RV32-NEXT:  # %bb.3:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_4:
-; RV32-NEXT:    fsd ft1, 120(sp)
+; RV32-NEXT:    fsd ft0, 120(sp)
+; RV32-NEXT:    bnez a0, .LBB23_7
+; RV32-NEXT:  # %bb.6:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 14
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_8
+; RV32-NEXT:  .LBB23_7:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 14
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_6
-; RV32-NEXT:  # %bb.5:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_6:
-; RV32-NEXT:    fsd ft1, 112(sp)
+; RV32-NEXT:  .LBB23_8:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 112(sp)
+; RV32-NEXT:    bnez a0, .LBB23_10
+; RV32-NEXT:  # %bb.9:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 13
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_11
+; RV32-NEXT:  .LBB23_10:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 13
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_8
-; RV32-NEXT:  # %bb.7:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_8:
-; RV32-NEXT:    fsd ft1, 104(sp)
+; RV32-NEXT:  .LBB23_11:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 104(sp)
+; RV32-NEXT:    bnez a0, .LBB23_13
+; RV32-NEXT:  # %bb.12:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 12
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_14
+; RV32-NEXT:  .LBB23_13:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 12
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_10
-; RV32-NEXT:  # %bb.9:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_10:
-; RV32-NEXT:    fsd ft1, 96(sp)
+; RV32-NEXT:  .LBB23_14:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 96(sp)
+; RV32-NEXT:    bnez a0, .LBB23_16
+; RV32-NEXT:  # %bb.15:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 11
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_17
+; RV32-NEXT:  .LBB23_16:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 11
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_12
-; RV32-NEXT:  # %bb.11:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_12:
-; RV32-NEXT:    fsd ft1, 88(sp)
+; RV32-NEXT:  .LBB23_17:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 88(sp)
+; RV32-NEXT:    bnez a0, .LBB23_19
+; RV32-NEXT:  # %bb.18:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 10
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_20
+; RV32-NEXT:  .LBB23_19:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 10
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_14
-; RV32-NEXT:  # %bb.13:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_14:
-; RV32-NEXT:    fsd ft1, 80(sp)
+; RV32-NEXT:  .LBB23_20:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 80(sp)
+; RV32-NEXT:    bnez a0, .LBB23_22
+; RV32-NEXT:  # %bb.21:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 9
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_23
+; RV32-NEXT:  .LBB23_22:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 9
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_16
-; RV32-NEXT:  # %bb.15:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_16:
-; RV32-NEXT:    fsd ft1, 72(sp)
+; RV32-NEXT:  .LBB23_23:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 72(sp)
+; RV32-NEXT:    bnez a0, .LBB23_25
+; RV32-NEXT:  # %bb.24:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 8
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_26
+; RV32-NEXT:  .LBB23_25:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 8
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_18
-; RV32-NEXT:  # %bb.17:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_18:
-; RV32-NEXT:    fsd ft1, 64(sp)
+; RV32-NEXT:  .LBB23_26:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 64(sp)
+; RV32-NEXT:    bnez a0, .LBB23_28
+; RV32-NEXT:  # %bb.27:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 7
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_29
+; RV32-NEXT:  .LBB23_28:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 7
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_20
-; RV32-NEXT:  # %bb.19:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_20:
-; RV32-NEXT:    fsd ft1, 56(sp)
+; RV32-NEXT:  .LBB23_29:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 56(sp)
+; RV32-NEXT:    bnez a0, .LBB23_31
+; RV32-NEXT:  # %bb.30:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 6
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_32
+; RV32-NEXT:  .LBB23_31:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 6
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_22
-; RV32-NEXT:  # %bb.21:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_22:
-; RV32-NEXT:    fsd ft1, 48(sp)
+; RV32-NEXT:  .LBB23_32:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 48(sp)
+; RV32-NEXT:    bnez a0, .LBB23_34
+; RV32-NEXT:  # %bb.33:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 5
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_35
+; RV32-NEXT:  .LBB23_34:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 5
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_24
-; RV32-NEXT:  # %bb.23:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_24:
-; RV32-NEXT:    fsd ft1, 40(sp)
+; RV32-NEXT:  .LBB23_35:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 40(sp)
+; RV32-NEXT:    bnez a0, .LBB23_37
+; RV32-NEXT:  # %bb.36:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 4
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_38
+; RV32-NEXT:  .LBB23_37:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 4
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_26
-; RV32-NEXT:  # %bb.25:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_26:
-; RV32-NEXT:    fsd ft1, 32(sp)
+; RV32-NEXT:  .LBB23_38:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 32(sp)
+; RV32-NEXT:    bnez a0, .LBB23_40
+; RV32-NEXT:  # %bb.39:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 3
-; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    j .LBB23_41
+; RV32-NEXT:  .LBB23_40:
+; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v8, 3
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_28
-; RV32-NEXT:  # %bb.27:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_28:
-; RV32-NEXT:    fsd ft1, 24(sp)
+; RV32-NEXT:  .LBB23_41:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 24(sp)
+; RV32-NEXT:    bnez a0, .LBB23_43
+; RV32-NEXT:  # %bb.42:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v24, v16, 2
-; RV32-NEXT:    vfmv.f.s ft0, v24
-; RV32-NEXT:    vslidedown.vi v24, v8, 2
-; RV32-NEXT:    vfmv.f.s ft1, v24
-; RV32-NEXT:    bnez a0, .LBB23_30
-; RV32-NEXT:  # %bb.29:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_30:
-; RV32-NEXT:    fsd ft1, 16(sp)
+; RV32-NEXT:    j .LBB23_44
+; RV32-NEXT:  .LBB23_43:
 ; RV32-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV32-NEXT:    vslidedown.vi v16, v16, 1
-; RV32-NEXT:    vfmv.f.s ft0, v16
+; RV32-NEXT:    vslidedown.vi v24, v8, 2
+; RV32-NEXT:  .LBB23_44:
+; RV32-NEXT:    vfmv.f.s ft0, v24
+; RV32-NEXT:    fsd ft0, 16(sp)
+; RV32-NEXT:    bnez a0, .LBB23_46
+; RV32-NEXT:  # %bb.45:
+; RV32-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
+; RV32-NEXT:    vslidedown.vi v8, v16, 1
+; RV32-NEXT:    j .LBB23_47
+; RV32-NEXT:  .LBB23_46:
+; RV32-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
 ; RV32-NEXT:    vslidedown.vi v8, v8, 1
-; RV32-NEXT:    vfmv.f.s ft1, v8
-; RV32-NEXT:    bnez a0, .LBB23_32
-; RV32-NEXT:  # %bb.31:
-; RV32-NEXT:    fmv.d ft1, ft0
-; RV32-NEXT:  .LBB23_32:
-; RV32-NEXT:    fsd ft1, 8(sp)
+; RV32-NEXT:  .LBB23_47:
+; RV32-NEXT:    vfmv.f.s ft0, v8
+; RV32-NEXT:    fsd ft0, 8(sp)
 ; RV32-NEXT:    vsetivli a0, 16, e64,m8,ta,mu
 ; RV32-NEXT:    vle64.v v8, (sp)
 ; RV32-NEXT:    addi sp, s0, -256
@@ -3581,164 +3916,181 @@
 ; RV64-NEXT:    .cfi_def_cfa s0, 0
 ; RV64-NEXT:    andi sp, sp, -128
 ; RV64-NEXT:    feq.d a0, fa0, fa1
-; RV64-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
-; RV64-NEXT:    vfmv.f.s ft1, v16
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    bnez a0, .LBB23_2
+; RV64-NEXT:    bnez a0, .LBB23_3
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    fmv.d ft0, ft1
+; RV64-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v16
+; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    beqz a0, .LBB23_4
 ; RV64-NEXT:  .LBB23_2:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
+; RV64-NEXT:    vslidedown.vi v24, v8, 15
+; RV64-NEXT:    j .LBB23_5
+; RV64-NEXT:  .LBB23_3:
+; RV64-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; RV64-NEXT:    vfmv.f.s ft0, v8
 ; RV64-NEXT:    fsd ft0, 0(sp)
+; RV64-NEXT:    bnez a0, .LBB23_2
+; RV64-NEXT:  .LBB23_4:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 15
+; RV64-NEXT:  .LBB23_5:
 ; RV64-NEXT:    vfmv.f.s ft0, v24
-; RV64-NEXT:    vslidedown.vi v24, v8, 15
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_4
-; RV64-NEXT:  # %bb.3:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_4:
-; RV64-NEXT:    fsd ft1, 120(sp)
+; RV64-NEXT:    fsd ft0, 120(sp)
+; RV64-NEXT:    bnez a0, .LBB23_7
+; RV64-NEXT:  # %bb.6:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 14
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_8
+; RV64-NEXT:  .LBB23_7:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 14
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_6
-; RV64-NEXT:  # %bb.5:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_6:
-; RV64-NEXT:    fsd ft1, 112(sp)
+; RV64-NEXT:  .LBB23_8:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 112(sp)
+; RV64-NEXT:    bnez a0, .LBB23_10
+; RV64-NEXT:  # %bb.9:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 13
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_11
+; RV64-NEXT:  .LBB23_10:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 13
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_8
-; RV64-NEXT:  # %bb.7:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_8:
-; RV64-NEXT:    fsd ft1, 104(sp)
+; RV64-NEXT:  .LBB23_11:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 104(sp)
+; RV64-NEXT:    bnez a0, .LBB23_13
+; RV64-NEXT:  # %bb.12:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 12
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_14
+; RV64-NEXT:  .LBB23_13:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 12
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_10
-; RV64-NEXT:  # %bb.9:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_10:
-; RV64-NEXT:    fsd ft1, 96(sp)
+; RV64-NEXT:  .LBB23_14:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 96(sp)
+; RV64-NEXT:    bnez a0, .LBB23_16
+; RV64-NEXT:  # %bb.15:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 11
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_17
+; RV64-NEXT:  .LBB23_16:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 11
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_12
-; RV64-NEXT:  # %bb.11:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_12:
-; RV64-NEXT:    fsd ft1, 88(sp)
+; RV64-NEXT:  .LBB23_17:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 88(sp)
+; RV64-NEXT:    bnez a0, .LBB23_19
+; RV64-NEXT:  # %bb.18:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 10
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_20
+; RV64-NEXT:  .LBB23_19:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 10
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_14
-; RV64-NEXT:  # %bb.13:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_14:
-; RV64-NEXT:    fsd ft1, 80(sp)
+; RV64-NEXT:  .LBB23_20:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 80(sp)
+; RV64-NEXT:    bnez a0, .LBB23_22
+; RV64-NEXT:  # %bb.21:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 9
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_23
+; RV64-NEXT:  .LBB23_22:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 9
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_16
-; RV64-NEXT:  # %bb.15:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_16:
-; RV64-NEXT:    fsd ft1, 72(sp)
+; RV64-NEXT:  .LBB23_23:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 72(sp)
+; RV64-NEXT:    bnez a0, .LBB23_25
+; RV64-NEXT:  # %bb.24:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 8
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_26
+; RV64-NEXT:  .LBB23_25:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 8
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_18
-; RV64-NEXT:  # %bb.17:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_18:
-; RV64-NEXT:    fsd ft1, 64(sp)
+; RV64-NEXT:  .LBB23_26:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 64(sp)
+; RV64-NEXT:    bnez a0, .LBB23_28
+; RV64-NEXT:  # %bb.27:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 7
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_29
+; RV64-NEXT:  .LBB23_28:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 7
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_20
-; RV64-NEXT:  # %bb.19:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_20:
-; RV64-NEXT:    fsd ft1, 56(sp)
+; RV64-NEXT:  .LBB23_29:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 56(sp)
+; RV64-NEXT:    bnez a0, .LBB23_31
+; RV64-NEXT:  # %bb.30:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 6
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_32
+; RV64-NEXT:  .LBB23_31:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 6
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_22
-; RV64-NEXT:  # %bb.21:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_22:
-; RV64-NEXT:    fsd ft1, 48(sp)
+; RV64-NEXT:  .LBB23_32:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 48(sp)
+; RV64-NEXT:    bnez a0, .LBB23_34
+; RV64-NEXT:  # %bb.33:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 5
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_35
+; RV64-NEXT:  .LBB23_34:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 5
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_24
-; RV64-NEXT:  # %bb.23:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_24:
-; RV64-NEXT:    fsd ft1, 40(sp)
+; RV64-NEXT:  .LBB23_35:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 40(sp)
+; RV64-NEXT:    bnez a0, .LBB23_37
+; RV64-NEXT:  # %bb.36:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 4
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_38
+; RV64-NEXT:  .LBB23_37:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 4
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_26
-; RV64-NEXT:  # %bb.25:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_26:
-; RV64-NEXT:    fsd ft1, 32(sp)
+; RV64-NEXT:  .LBB23_38:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 32(sp)
+; RV64-NEXT:    bnez a0, .LBB23_40
+; RV64-NEXT:  # %bb.39:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 3
-; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    j .LBB23_41
+; RV64-NEXT:  .LBB23_40:
+; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v8, 3
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_28
-; RV64-NEXT:  # %bb.27:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_28:
-; RV64-NEXT:    fsd ft1, 24(sp)
+; RV64-NEXT:  .LBB23_41:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 24(sp)
+; RV64-NEXT:    bnez a0, .LBB23_43
+; RV64-NEXT:  # %bb.42:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v24, v16, 2
-; RV64-NEXT:    vfmv.f.s ft0, v24
-; RV64-NEXT:    vslidedown.vi v24, v8, 2
-; RV64-NEXT:    vfmv.f.s ft1, v24
-; RV64-NEXT:    bnez a0, .LBB23_30
-; RV64-NEXT:  # %bb.29:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_30:
-; RV64-NEXT:    fsd ft1, 16(sp)
+; RV64-NEXT:    j .LBB23_44
+; RV64-NEXT:  .LBB23_43:
 ; RV64-NEXT:    vsetivli a1, 1, e64,m8,ta,mu
-; RV64-NEXT:    vslidedown.vi v16, v16, 1
-; RV64-NEXT:    vfmv.f.s ft0, v16
+; RV64-NEXT:    vslidedown.vi v24, v8, 2
+; RV64-NEXT:  .LBB23_44:
+; RV64-NEXT:    vfmv.f.s ft0, v24
+; RV64-NEXT:    fsd ft0, 16(sp)
+; RV64-NEXT:    bnez a0, .LBB23_46
+; RV64-NEXT:  # %bb.45:
+; RV64-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
+; RV64-NEXT:    vslidedown.vi v8, v16, 1
+; RV64-NEXT:    j .LBB23_47
+; RV64-NEXT:  .LBB23_46:
+; RV64-NEXT:    vsetivli a0, 1, e64,m8,ta,mu
 ; RV64-NEXT:    vslidedown.vi v8, v8, 1
-; RV64-NEXT:    vfmv.f.s ft1, v8
-; RV64-NEXT:    bnez a0, .LBB23_32
-; RV64-NEXT:  # %bb.31:
-; RV64-NEXT:    fmv.d ft1, ft0
-; RV64-NEXT:  .LBB23_32:
-; RV64-NEXT:    fsd ft1, 8(sp)
+; RV64-NEXT:  .LBB23_47:
+; RV64-NEXT:    vfmv.f.s ft0, v8
+; RV64-NEXT:    fsd ft0, 8(sp)
 ; RV64-NEXT:    vsetivli a0, 16, e64,m8,ta,mu
 ; RV64-NEXT:    vle64.v v8, (sp)
 ; RV64-NEXT:    addi sp, s0, -256
diff --git a/llvm/test/CodeGen/RISCV/rvv/frameindex-addr.ll b/llvm/test/CodeGen/RISCV/rvv/frameindex-addr.ll
--- a/llvm/test/CodeGen/RISCV/rvv/frameindex-addr.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/frameindex-addr.ll
@@ -15,8 +15,7 @@
   ; CHECK: bb.0.entry:
   ; CHECK:   liveins: $v8
   ; CHECK:   [[COPY:%[0-9]+]]:vr = COPY $v8
-  ; CHECK:   dead %2:gpr = PseudoVSETIVLI 1, 88, implicit-def $vl, implicit-def $vtype
-  ; CHECK:   PseudoVSE64_V_M1 [[COPY]], %stack.0.a, 1, 6, implicit $vl, implicit $vtype
+  ; CHECK:   PseudoVSE64_V_M1 [[COPY]], %stack.0.a, 1, 6
   ; CHECK:   [[LD:%[0-9]+]]:gpr = LD %stack.0.a, 0 :: (dereferenceable load 8 from %ir.a)
   ; CHECK:   $x10 = COPY [[LD]]
   ; CHECK:   PseudoRET implicit $x10
diff --git a/llvm/test/CodeGen/RISCV/rvv/mask-reg-alloc.mir b/llvm/test/CodeGen/RISCV/rvv/mask-reg-alloc.mir
--- a/llvm/test/CodeGen/RISCV/rvv/mask-reg-alloc.mir
+++ b/llvm/test/CodeGen/RISCV/rvv/mask-reg-alloc.mir
@@ -15,20 +15,21 @@
     liveins: $v0, $v1, $v2, $v3
     ; CHECK-LABEL: name: mask_reg_alloc
     ; CHECK: liveins: $v0, $v1, $v2, $v3
-    ; CHECK: renamable $v25 = PseudoVMERGE_VIM_M1 killed renamable $v2, 1, killed renamable $v0, $noreg, -1, implicit $vl, implicit $vtype
+    ; CHECK: dead renamable $x10 = PseudoVSETIVLI 1, 64, implicit-def $vl, implicit-def $vtype
+    ; CHECK: renamable $v25 = PseudoVMERGE_VIM_M1 killed renamable $v2, 1, killed renamable $v0, 1, 3, implicit $vl, implicit $vtype
     ; CHECK: renamable $v0 = COPY killed renamable $v1
-    ; CHECK: renamable $v26 = PseudoVMERGE_VIM_M1 killed renamable $v3, 1, killed renamable $v0, $noreg, -1, implicit $vl, implicit $vtype
-    ; CHECK: renamable $v0 = PseudoVADD_VV_M1 killed renamable $v25, killed renamable $v26, $noreg, -1, implicit $vl, implicit $vtype
+    ; CHECK: renamable $v26 = PseudoVMERGE_VIM_M1 killed renamable $v3, 1, killed renamable $v0, 1, 3, implicit $vl, implicit $vtype
+    ; CHECK: renamable $v0 = PseudoVADD_VV_M1 killed renamable $v25, killed renamable $v26, 1, 3, implicit $vl, implicit $vtype
     ; CHECK: PseudoRET implicit $v0
     %0:vr = COPY $v0
     %1:vr = COPY $v1
     %2:vr = COPY $v2
     %3:vr = COPY $v3
     %4:vmv0 = COPY %0
-    %5:vrnov0 = PseudoVMERGE_VIM_M1 killed %2, 1, %4, $noreg, -1, implicit $vl, implicit $vtype
+    %5:vrnov0 = PseudoVMERGE_VIM_M1 killed %2, 1, %4, 1, 3
     %6:vmv0 = COPY %1
-    %7:vrnov0 = PseudoVMERGE_VIM_M1 killed %3, 1, %6, $noreg, -1, implicit $vl, implicit $vtype
-    %8:vr = PseudoVADD_VV_M1 killed %5, killed %7, $noreg, -1, implicit $vl, implicit $vtype
+    %7:vrnov0 = PseudoVMERGE_VIM_M1 killed %3, 1, %6, 1, 3
+    %8:vr = PseudoVADD_VV_M1 killed %5, killed %7, 1, 3
     $v0 = COPY %8
     PseudoRET implicit $v0
 ...
diff --git a/llvm/test/CodeGen/RISCV/rvv/tail-agnostic-impdef-copy.mir b/llvm/test/CodeGen/RISCV/rvv/tail-agnostic-impdef-copy.mir
--- a/llvm/test/CodeGen/RISCV/rvv/tail-agnostic-impdef-copy.mir
+++ b/llvm/test/CodeGen/RISCV/rvv/tail-agnostic-impdef-copy.mir
@@ -52,8 +52,7 @@
     ; CHECK: $v0 = COPY [[COPY]]
     ; CHECK: [[DEF:%[0-9]+]]:vrm8 = IMPLICIT_DEF
     ; CHECK: [[COPY2:%[0-9]+]]:vrm8nov0 = COPY [[DEF]]
-    ; CHECK: dead %5:gpr = PseudoVSETVLI $x0, 91, implicit-def $vl, implicit-def $vtype
-    ; CHECK: [[PseudoVLE64_V_M8_MASK:%[0-9]+]]:vrm8nov0 = PseudoVLE64_V_M8_MASK [[COPY2]], [[COPY1]], $v0, $noreg, 6, implicit $vl, implicit $vtype :: (load 64 from %ir.a, align 8)
+    ; CHECK: [[PseudoVLE64_V_M8_MASK:%[0-9]+]]:vrm8nov0 = PseudoVLE64_V_M8_MASK [[COPY2]], [[COPY1]], $v0, $x0, 6, implicit $vl, implicit $vtype :: (load 64 from %ir.a, align 8)
     ; CHECK: $v8m8 = COPY [[PseudoVLE64_V_M8_MASK]]
     ; CHECK: PseudoRET implicit $v8m8
     %1:vr = COPY $v0