[RISCV] Prune more unnecessary vector pseudo instructions. NFC

[craig.topper]

Summary

For floating point specific vector instructions, we don't need
pseudos for mf8.

[khchen]

LGTM.

llvm/lib/Target/RISCV/RISCVInstrInfo.cpp

bool RISCVInstrInfo::findCommutedOpIndices(const MachineInstr &MI,
   case CASE_VFMA_SPLATS(FMADD):
   case CASE_VFMA_SPLATS(FMSUB):
   case CASE_VFMA_SPLATS(FMACC):
   case CASE_VFMA_SPLATS(FMSAC):
   case CASE_VFMA_SPLATS(FNMADD):
   case CASE_VFMA_SPLATS(FNMSUB):
   case CASE_VFMA_SPLATS(FNMACC):
   case CASE_VFMA_SPLATS(FNMSAC):
-  case CASE_VFMA_OPCODE_LMULS(FMACC, VV):
-  case CASE_VFMA_OPCODE_LMULS(FMSAC, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMACC, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMSAC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMACC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMSAC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMACC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMSAC, VV):
   case CASE_VFMA_OPCODE_LMULS(MADD, VX):
   case CASE_VFMA_OPCODE_LMULS(NMSUB, VX):
   case CASE_VFMA_OPCODE_LMULS(MACC, VX):
   case CASE_VFMA_OPCODE_LMULS(NMSAC, VX):
   case CASE_VFMA_OPCODE_LMULS(MACC, VV):
   case CASE_VFMA_OPCODE_LMULS(NMSAC, VV): {
     // If the tail policy is undisturbed we can't commute.
     assert(RISCVII::hasVecPolicyOp(MI.getDesc().TSFlags));
     if ((MI.getOperand(MI.getNumExplicitOperands() - 1).getImm() & 1) == 0)
       return false;
 
     // For these instructions we can only swap operand 1 and operand 3 by
     // changing the opcode.
     unsigned CommutableOpIdx1 = 1;
     unsigned CommutableOpIdx2 = 3;
     if (!fixCommutedOpIndices(SrcOpIdx1, SrcOpIdx2, CommutableOpIdx1,
                               CommutableOpIdx2))
       return false;
     return true;
   }
-  case CASE_VFMA_OPCODE_LMULS(FMADD, VV):
-  case CASE_VFMA_OPCODE_LMULS(FMSUB, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMADD, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMSUB, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMADD, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMSUB, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMADD, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMSUB, VV):
   case CASE_VFMA_OPCODE_LMULS(MADD, VV):
   case CASE_VFMA_OPCODE_LMULS(NMSUB, VV): {
     // If the tail policy is undisturbed we can't commute.
     assert(RISCVII::hasVecPolicyOp(MI.getDesc().TSFlags));
     if ((MI.getOperand(MI.getNumExplicitOperands() - 1).getImm() & 1) == 0)
       return false;
 
     // For these instructions we have more freedom. We can commute with the
MachineInstr *RISCVInstrInfo::commuteInstructionImpl(MachineInstr &MI,
   case CASE_VFMA_SPLATS(FMACC):
   case CASE_VFMA_SPLATS(FMADD):
   case CASE_VFMA_SPLATS(FMSAC):
   case CASE_VFMA_SPLATS(FMSUB):
   case CASE_VFMA_SPLATS(FNMACC):
   case CASE_VFMA_SPLATS(FNMADD):
   case CASE_VFMA_SPLATS(FNMSAC):
   case CASE_VFMA_SPLATS(FNMSUB):
-  case CASE_VFMA_OPCODE_LMULS(FMACC, VV):
-  case CASE_VFMA_OPCODE_LMULS(FMSAC, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMACC, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMSAC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMACC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMSAC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMACC, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMSAC, VV):
   case CASE_VFMA_OPCODE_LMULS(MADD, VX):
   case CASE_VFMA_OPCODE_LMULS(NMSUB, VX):
   case CASE_VFMA_OPCODE_LMULS(MACC, VX):
   case CASE_VFMA_OPCODE_LMULS(NMSAC, VX):
   case CASE_VFMA_OPCODE_LMULS(MACC, VV):
   case CASE_VFMA_OPCODE_LMULS(NMSAC, VV): {
     // It only make sense to toggle these between clobbering the
     // addend/subtrahend/minuend one of the multiplicands.
     assert((OpIdx1 == 1 || OpIdx2 == 1) && "Unexpected opcode index");
     assert((OpIdx1 == 3 || OpIdx2 == 3) && "Unexpected opcode index");
     unsigned Opc;
     switch (MI.getOpcode()) {
       default:
         llvm_unreachable("Unexpected opcode");
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FMACC, FMADD)
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FMADD, FMACC)
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FMSAC, FMSUB)
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FMSUB, FMSAC)
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FNMACC, FNMADD)
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FNMADD, FNMACC)
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FNMSAC, FNMSUB)
       CASE_VFMA_CHANGE_OPCODE_SPLATS(FNMSUB, FNMSAC)
-      CASE_VFMA_CHANGE_OPCODE_LMULS(FMACC, FMADD, VV)
-      CASE_VFMA_CHANGE_OPCODE_LMULS(FMSAC, FMSUB, VV)
-      CASE_VFMA_CHANGE_OPCODE_LMULS(FNMACC, FNMADD, VV)
-      CASE_VFMA_CHANGE_OPCODE_LMULS(FNMSAC, FNMSUB, VV)
+      CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FMACC, FMADD, VV)
+      CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FMSAC, FMSUB, VV)
+      CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FNMACC, FNMADD, VV)
+      CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FNMSAC, FNMSUB, VV)
       CASE_VFMA_CHANGE_OPCODE_LMULS(MACC, MADD, VX)
       CASE_VFMA_CHANGE_OPCODE_LMULS(MADD, MACC, VX)
       CASE_VFMA_CHANGE_OPCODE_LMULS(NMSAC, NMSUB, VX)
       CASE_VFMA_CHANGE_OPCODE_LMULS(NMSUB, NMSAC, VX)
       CASE_VFMA_CHANGE_OPCODE_LMULS(MACC, MADD, VV)
       CASE_VFMA_CHANGE_OPCODE_LMULS(NMSAC, NMSUB, VV)
     }
 
     auto &WorkingMI = cloneIfNew(MI);
     WorkingMI.setDesc(get(Opc));
     return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
                                                    OpIdx1, OpIdx2);
   }
-  case CASE_VFMA_OPCODE_LMULS(FMADD, VV):
-  case CASE_VFMA_OPCODE_LMULS(FMSUB, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMADD, VV):
-  case CASE_VFMA_OPCODE_LMULS(FNMSUB, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMADD, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FMSUB, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMADD, VV):
+  case CASE_VFMA_OPCODE_LMULS_MF4(FNMSUB, VV):
   case CASE_VFMA_OPCODE_LMULS(MADD, VV):
   case CASE_VFMA_OPCODE_LMULS(NMSUB, VV): {
     assert((OpIdx1 == 1 || OpIdx2 == 1) && "Unexpected opcode index");
     // If one of the operands, is the addend we need to change opcode.
     // Otherwise we're just swapping 2 of the multiplicands.
     if (OpIdx1 == 3 || OpIdx2 == 3) {
       unsigned Opc;
       switch (MI.getOpcode()) {
         default:
           llvm_unreachable("Unexpected opcode");
-        CASE_VFMA_CHANGE_OPCODE_LMULS(FMADD, FMACC, VV)
-        CASE_VFMA_CHANGE_OPCODE_LMULS(FMSUB, FMSAC, VV)
-        CASE_VFMA_CHANGE_OPCODE_LMULS(FNMADD, FNMACC, VV)
-        CASE_VFMA_CHANGE_OPCODE_LMULS(FNMSUB, FNMSAC, VV)
+        CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FMADD, FMACC, VV)
+        CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FMSUB, FMSAC, VV)
+        CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FNMADD, FNMACC, VV)
+        CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(FNMSUB, FNMSAC, VV)
         CASE_VFMA_CHANGE_OPCODE_LMULS(MADD, MACC, VV)
         CASE_VFMA_CHANGE_OPCODE_LMULS(NMSUB, NMSAC, VV)
       }
 
       auto &WorkingMI = cloneIfNew(MI);
       WorkingMI.setDesc(get(Opc));
       return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
                                                      OpIdx1, OpIdx2);

 #undef CASE_VFMA_SPLATS
 #undef CASE_VFMA_OPCODE_LMULS
 #undef CASE_VFMA_OPCODE_COMMON
 
 // clang-format off
 #define CASE_WIDEOP_OPCODE_COMMON(OP, LMUL)                                    \
   RISCV::PseudoV##OP##_##LMUL##_TIED
 
-#define CASE_WIDEOP_OPCODE_LMULS(OP)                                           \
-  CASE_WIDEOP_OPCODE_COMMON(OP, MF8):                                          \
-  case CASE_WIDEOP_OPCODE_COMMON(OP, MF4):                                     \
+#define CASE_WIDEOP_OPCODE_LMULS_MF4(OP)                                       \
+  CASE_WIDEOP_OPCODE_COMMON(OP, MF4):                                          \
   case CASE_WIDEOP_OPCODE_COMMON(OP, MF2):                                     \
   case CASE_WIDEOP_OPCODE_COMMON(OP, M1):                                      \
   case CASE_WIDEOP_OPCODE_COMMON(OP, M2):                                      \
   case CASE_WIDEOP_OPCODE_COMMON(OP, M4)
+
+#define CASE_WIDEOP_OPCODE_LMULS(OP)                                           \
+  CASE_WIDEOP_OPCODE_COMMON(OP, MF8):                                          \
+  case CASE_WIDEOP_OPCODE_LMULS_MF4(OP)
 // clang-format on
 
 #define CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, LMUL)                             \
   case RISCV::PseudoV##OP##_##LMUL##_TIED:                                     \
     NewOpc = RISCV::PseudoV##OP##_##LMUL;                                      \
     break;
 
-#define CASE_WIDEOP_CHANGE_OPCODE_LMULS(OP)                                    \
-  CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, MF8)                                    \
+#define CASE_WIDEOP_CHANGE_OPCODE_LMULS_MF4(OP)                                 \
   CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, MF4)                                    \
   CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, MF2)                                    \
   CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, M1)                                     \
   CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, M2)                                     \
   CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, M4)
 
+#define CASE_WIDEOP_CHANGE_OPCODE_LMULS(OP)                                    \
+  CASE_WIDEOP_CHANGE_OPCODE_COMMON(OP, MF8)                                    \
+  CASE_WIDEOP_CHANGE_OPCODE_LMULS_MF4(OP)
+
 MachineInstr *RISCVInstrInfo::convertToThreeAddress(MachineInstr &MI,
                                                     LiveVariables *LV,
                                                     LiveIntervals *LIS) const {
   switch (MI.getOpcode()) {
   default:
     break;
-  case CASE_WIDEOP_OPCODE_LMULS(FWADD_WV):
-  case CASE_WIDEOP_OPCODE_LMULS(FWSUB_WV):
+  case CASE_WIDEOP_OPCODE_LMULS_MF4(FWADD_WV):
+  case CASE_WIDEOP_OPCODE_LMULS_MF4(FWSUB_WV):
   case CASE_WIDEOP_OPCODE_LMULS(WADD_WV):
   case CASE_WIDEOP_OPCODE_LMULS(WADDU_WV):
   case CASE_WIDEOP_OPCODE_LMULS(WSUB_WV):
   case CASE_WIDEOP_OPCODE_LMULS(WSUBU_WV): {
     // clang-format off
     unsigned NewOpc;
     switch (MI.getOpcode()) {
     default:
       llvm_unreachable("Unexpected opcode");
-    CASE_WIDEOP_CHANGE_OPCODE_LMULS(FWADD_WV)
-    CASE_WIDEOP_CHANGE_OPCODE_LMULS(FWSUB_WV)
+    CASE_WIDEOP_CHANGE_OPCODE_LMULS_MF4(FWADD_WV)
+    CASE_WIDEOP_CHANGE_OPCODE_LMULS_MF4(FWSUB_WV)
     CASE_WIDEOP_CHANGE_OPCODE_LMULS(WADD_WV)
     CASE_WIDEOP_CHANGE_OPCODE_LMULS(WADDU_WV)
     CASE_WIDEOP_CHANGE_OPCODE_LMULS(WSUB_WV)
     CASE_WIDEOP_CHANGE_OPCODE_LMULS(WSUBU_WV)
     }
     //clang-format on
 
     MachineBasicBlock &MBB = *MI.getParent();

llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td

 def V_M8  : LMULInfo<0b011, 64, VRM8,/*NoVReg*/VR, VRM4, VRM2, VR, "M8">;
 
 def V_MF8 : LMULInfo<0b101, 1, VR, VR,/*NoVReg*/VR,/*NoVReg*/VR,/*NoVReg*/VR, "MF8">;
 def V_MF4 : LMULInfo<0b110, 2, VR, VR,          VR,/*NoVReg*/VR,/*NoVReg*/VR, "MF4">;
 def V_MF2 : LMULInfo<0b111, 4, VR, VR,          VR,          VR,/*NoVReg*/VR, "MF2">;
 
 // Used to iterate over all possible LMULs.
 defvar MxList = [V_MF8, V_MF4, V_MF2, V_M1, V_M2, V_M4, V_M8];
+// For floating point which don't need MF8.
+defvar MxListF = [V_MF4, V_MF2, V_M1, V_M2, V_M4, V_M8];
 
 // Used for widening and narrowing instructions as it doesn't contain M8.
 defvar MxListW = [V_MF8, V_MF4, V_MF2, V_M1, V_M2, V_M4];
+// For floating point which don't need MF8.
+defvar MxListFW = [V_MF4, V_MF2, V_M1, V_M2, V_M4];
 
 // Use for zext/sext.vf2
 defvar MxListVF2 = [V_MF4, V_MF2, V_M1, V_M2, V_M4, V_M8];
 
 // Use for zext/sext.vf4
 defvar MxListVF4 = [V_MF2, V_M1, V_M2, V_M4, V_M8];
 
 // Use for zext/sext.vf8
multiclass VPseudoTiedBinary<VReg RetClass,
   }
 }
 
 multiclass VPseudoBinaryV_VV<string Constraint = ""> {
   foreach m = MxList in
     defm _VV : VPseudoBinary<m.vrclass, m.vrclass, m.vrclass, m, Constraint>;
 }
 
+// Similar to VPseudoBinaryV_VV, but uses MxListF.
+multiclass VPseudoBinaryFV_VV<string Constraint = ""> {
+  foreach m = MxListF in
+    defm _VV : VPseudoBinary<m.vrclass, m.vrclass, m.vrclass, m, Constraint>;
+}
+
 multiclass VPseudoVGTR_VV_EEW<int eew, string Constraint = ""> {
   foreach m = MxList in {
     foreach sew = EEWList in {
       defvar octuple_lmul = m.octuple;
       // emul = lmul * eew / sew
       defvar octuple_emul = !srl(!mul(octuple_lmul, eew), log2<sew>.val);
       if !and(!ge(octuple_emul, 1), !le(octuple_emul, 64)) then {
         defvar emulMX = octuple_to_str<octuple_emul>.ret;

 
 // We use earlyclobber here due to
 // * The destination EEW is smaller than the source EEW and the overlap is
 //   in the lowest-numbered part of the source register group is legal.
 //   Otherwise, it is illegal.
 // * The destination EEW is greater than the source EEW, the source EMUL is
 //   at least 1, and the overlap is in the highest-numbered part of the
 //   destination register group is legal. Otherwise, it is illegal.
-multiclass VPseudoBinaryW_VV {
-  foreach m = MxListW in
+multiclass VPseudoBinaryW_VV<list<LMULInfo> mxlist = MxListW> {
+  foreach m = mxlist in
     defm _VV : VPseudoBinary<m.wvrclass, m.vrclass, m.vrclass, m,
                              "@earlyclobber $rd">;
 }
 
 multiclass VPseudoBinaryW_VX {
   foreach m = MxListW in
     defm "_VX" : VPseudoBinary<m.wvrclass, m.vrclass, GPR, m,
                                "@earlyclobber $rd">;
 }
 
 multiclass VPseudoBinaryW_VF {
   foreach f = FPListW in
     foreach m = f.MxList in
       defm "_V" # f.FX : VPseudoBinary<m.wvrclass, m.vrclass,
                                        f.fprclass, m,
                                        "@earlyclobber $rd">;
 }
 
-multiclass VPseudoBinaryW_WV {
-  foreach m = MxListW in {
+multiclass VPseudoBinaryW_WV<list<LMULInfo> mxlist = MxListW> {
+  foreach m = mxlist in {
     defm _WV : VPseudoBinary<m.wvrclass, m.wvrclass, m.vrclass, m,
                              "@earlyclobber $rd">;
     defm _WV : VPseudoTiedBinary<m.wvrclass, m.vrclass, m,
                                  "@earlyclobber $rd">;
   }
 }
 
 multiclass VPseudoBinaryW_WX {
foreach m = f.MxList in {
           VPseudoUnaryNoDummyMask<m.vrclass, f.fprclass>,
           Sched<[WriteVFMovV, ReadVFMovF]>;
       }
     }
   }
 }
 
 multiclass VPseudoVCLS_V {
-  foreach m = MxList in {
+  foreach m = MxListF in {
     let VLMul = m.value in {
       def "_V_" # m.MX : VPseudoUnaryNoMask<m.vrclass, m.vrclass>,
                          Sched<[WriteVFClassV, ReadVFClassV, ReadVMask]>;
       def "_V_" # m.MX # "_MASK" : VPseudoUnaryMask<m.vrclass, m.vrclass>,
                                    Sched<[WriteVFClassV, ReadVFClassV, ReadVMask]>;
     }
   }
 }
 
 multiclass VPseudoVSQR_V {
-  foreach m = MxList in {
+  foreach m = MxListF in {
     let VLMul = m.value in {
       def "_V_" # m.MX : VPseudoUnaryNoMask<m.vrclass, m.vrclass>,
                          Sched<[WriteVFSqrtV, ReadVFSqrtV, ReadVMask]>;
       def "_V_" # m.MX # "_MASK" : VPseudoUnaryMaskTA<m.vrclass, m.vrclass>,
                                    Sched<[WriteVFSqrtV, ReadVFSqrtV, ReadVMask]>;
     }
   }
 }
 
 multiclass VPseudoVRCP_V {
-  foreach m = MxList in {
+  foreach m = MxListF in {
     let VLMul = m.value in {
       def "_V_" # m.MX : VPseudoUnaryNoMask<m.vrclass, m.vrclass>,
                          Sched<[WriteVFRecpV, ReadVFRecpV, ReadVMask]>;
       def "_V_" # m.MX # "_MASK" : VPseudoUnaryMaskTA<m.vrclass, m.vrclass>,
                                    Sched<[WriteVFRecpV, ReadVFRecpV, ReadVMask]>;
     }
   }
 }

 // When the destination EEW is different from source EEW, we need to use
 // @earlyclobber to avoid the overlap between destination and source registers.
 // We don't need @earlyclobber for LMUL<=1 since that matches this overlap
 // exception from the spec
 // "The destination EEW is smaller than the source EEW and the overlap is in the
 //  lowest-numbered part of the source register group".
 // With LMUL<=1 the source and dest occupy a single register so any overlap
 // is in the lowest-numbered part.
-multiclass VPseudoBinaryM_VV {
-  foreach m = MxList in
+multiclass VPseudoBinaryM_VV<list<LMULInfo> mxlist = MxList> {
+  foreach m = mxlist in
     defm _VV : VPseudoBinaryM<VR, m.vrclass, m.vrclass, m,
                               !if(!ge(m.octuple, 16), "@earlyclobber $rd", "")>;
 }
 
 multiclass VPseudoBinaryM_VX {
   foreach m = MxList in
     defm "_VX" :
       VPseudoBinaryM<VR, m.vrclass, GPR, m,

 multiclass VPseudoVDIV_VV_VX {
   defm "" : VPseudoBinaryV_VV,
             Sched<[WriteVIDivV, ReadVIDivV, ReadVIDivV, ReadVMask]>;
   defm "" : VPseudoBinaryV_VX,
             Sched<[WriteVIDivX, ReadVIDivV, ReadVIDivX, ReadVMask]>;
 }
 
 multiclass VPseudoVFMUL_VV_VF {
-  defm "" : VPseudoBinaryV_VV,
+  defm "" : VPseudoBinaryFV_VV,
             Sched<[WriteVFMulV, ReadVFMulV, ReadVFMulV, ReadVMask]>;
   defm "" : VPseudoBinaryV_VF,
             Sched<[WriteVFMulF, ReadVFMulV, ReadVFMulF, ReadVMask]>;
 }
 
 multiclass VPseudoVFDIV_VV_VF {
-  defm "" : VPseudoBinaryV_VV,
+  defm "" : VPseudoBinaryFV_VV,
             Sched<[WriteVFDivV, ReadVFDivV, ReadVFDivV, ReadVMask]>;
   defm "" : VPseudoBinaryV_VF,
             Sched<[WriteVFDivF, ReadVFDivV, ReadVFDivF, ReadVMask]>;
 }
 
 multiclass VPseudoVFRDIV_VF {
   defm "" : VPseudoBinaryV_VF,
             Sched<[WriteVFDivF, ReadVFDivV, ReadVFDivF, ReadVMask]>;
 }
 
 multiclass VPseudoVALU_VV_VX {
   defm "" : VPseudoBinaryV_VV,
             Sched<[WriteVIALUV, ReadVIALUV, ReadVIALUV, ReadVMask]>;
   defm "" : VPseudoBinaryV_VX,
             Sched<[WriteVIALUX, ReadVIALUV, ReadVIALUX, ReadVMask]>;
 }
 
 multiclass VPseudoVSGNJ_VV_VF {
-  defm "" : VPseudoBinaryV_VV,
+  defm "" : VPseudoBinaryFV_VV,
             Sched<[WriteVFSgnjV, ReadVFSgnjV, ReadVFSgnjV, ReadVMask]>;
   defm "" : VPseudoBinaryV_VF,
             Sched<[WriteVFSgnjF, ReadVFSgnjV, ReadVFSgnjF, ReadVMask]>;
 }
 
 multiclass VPseudoVMAX_VV_VF {
-  defm "" : VPseudoBinaryV_VV,
+  defm "" : VPseudoBinaryFV_VV,
             Sched<[WriteVFCmpV, ReadVFCmpV, ReadVFCmpV, ReadVMask]>;
   defm "" : VPseudoBinaryV_VF,
             Sched<[WriteVFCmpF, ReadVFCmpV, ReadVFCmpF, ReadVMask]>;
 }
 
 multiclass VPseudoVALU_VV_VF {
-  defm "" : VPseudoBinaryV_VV,
+  defm "" : VPseudoBinaryFV_VV,
             Sched<[WriteVFALUV, ReadVFALUV, ReadVFALUV, ReadVMask]>;
   defm "" : VPseudoBinaryV_VF,
             Sched<[WriteVFALUF, ReadVFALUV, ReadVFALUF, ReadVMask]>;
 }
 
 multiclass VPseudoVALU_VF {
   defm "" : VPseudoBinaryV_VF,
             Sched<[WriteVFALUF, ReadVFALUV, ReadVFALUF, ReadVMask]>;

 multiclass VPseudoVWMUL_VV_VX {
   defm "" : VPseudoBinaryW_VV,
             Sched<[WriteVIWMulV, ReadVIWMulV, ReadVIWMulV, ReadVMask]>;
   defm "" : VPseudoBinaryW_VX,
             Sched<[WriteVIWMulX, ReadVIWMulV, ReadVIWMulX, ReadVMask]>;
 }
 
 multiclass VPseudoVWMUL_VV_VF {
-  defm "" : VPseudoBinaryW_VV,
+  defm "" : VPseudoBinaryW_VV<MxListFW>,
             Sched<[WriteVFWMulV, ReadVFWMulV, ReadVFWMulV, ReadVMask]>;
   defm "" : VPseudoBinaryW_VF,
             Sched<[WriteVFWMulF, ReadVFWMulV, ReadVFWMulF, ReadVMask]>;
 }
 
-multiclass VPseudoBinaryW_VV_VF {
-  defm "" : VPseudoBinaryW_VV;
-  defm "" : VPseudoBinaryW_VF;
-}
-
 multiclass VPseudoVWALU_WV_WX {
   defm "" : VPseudoBinaryW_WV,
             Sched<[WriteVIWALUV, ReadVIWALUV, ReadVIWALUV, ReadVMask]>;
   defm "" : VPseudoBinaryW_WX,
             Sched<[WriteVIWALUX, ReadVIWALUV, ReadVIWALUX, ReadVMask]>;
 }
 
 multiclass VPseudoVFWALU_VV_VF {
-  defm "" : VPseudoBinaryW_VV,
+  defm "" : VPseudoBinaryW_VV<MxListFW>,
             Sched<[WriteVFWALUV, ReadVFWALUV, ReadVFWALUV, ReadVMask]>;
   defm "" : VPseudoBinaryW_VF,
             Sched<[WriteVFWALUF, ReadVFWALUV, ReadVFWALUF, ReadVMask]>;
 }
 
 multiclass VPseudoVFWALU_WV_WF {
-  defm "" : VPseudoBinaryW_WV,
+  defm "" : VPseudoBinaryW_WV<MxListFW>,
             Sched<[WriteVFWALUV, ReadVFWALUV, ReadVFWALUV, ReadVMask]>;
   defm "" : VPseudoBinaryW_WF,
             Sched<[WriteVFWALUF, ReadVFWALUV, ReadVFWALUF, ReadVMask]>;
 }
 
 multiclass VPseudoVMRG_VM_XM_IM {
   defm "" : VPseudoBinaryV_VM,
             Sched<[WriteVIMergeV, ReadVIMergeV, ReadVIMergeV, ReadVMask]>;
multiclass VPseudoTernaryWithPolicy<VReg RetClass,
                                     bit Commutable = 0> {
   let VLMul = MInfo.value in {
     let isCommutable = Commutable in
     def "_" # MInfo.MX : VPseudoTernaryNoMaskWithPolicy<RetClass, Op1Class, Op2Class, Constraint>;
     def "_" # MInfo.MX # "_MASK" : VPseudoBinaryMask<RetClass, Op1Class, Op2Class, Constraint>;
   }
 }
 
-multiclass VPseudoTernaryV_VV_AAXA<string Constraint = ""> {
-  foreach m = MxList in {
+multiclass VPseudoTernaryV_VV_AAXA<string Constraint = "",
+                                   list<LMULInfo> mxlist = MxList> {
+  foreach m = mxlist in {
     defm _VV : VPseudoTernaryWithPolicy<m.vrclass, m.vrclass, m.vrclass, m,
                                         Constraint, /*Commutable*/1>;
   }
 }
 
 multiclass VPseudoTernaryV_VX<string Constraint = ""> {
   foreach m = MxList in
     defm _VX : VPseudoTernary<m.vrclass, m.vrclass, GPR, m, Constraint>;
 }
 
 multiclass VPseudoTernaryV_VX_AAXA<string Constraint = ""> {
   foreach m = MxList in
     defm "_VX" : VPseudoTernaryWithPolicy<m.vrclass, GPR, m.vrclass, m,
                                           Constraint, /*Commutable*/1>;
 }
 
 multiclass VPseudoTernaryV_VF_AAXA<string Constraint = ""> {
   foreach f = FPList in
     foreach m = f.MxList in
       defm "_V" # f.FX : VPseudoTernaryWithPolicy<m.vrclass, f.fprclass,
                                                   m.vrclass, m, Constraint,
                                                   /*Commutable*/1>;
 }
 
-multiclass VPseudoTernaryW_VV {
+multiclass VPseudoTernaryW_VV<list<LMULInfo> mxlist = MxListW> {
   defvar constraint = "@earlyclobber $rd";
-  foreach m = MxListW in
+  foreach m = mxlist in
     defm _VV : VPseudoTernaryWithPolicy<m.wvrclass, m.vrclass, m.vrclass, m,
                                         constraint>;
 }
 
 multiclass VPseudoTernaryW_VX {
   defvar constraint = "@earlyclobber $rd";
   foreach m = MxListW in
     defm "_VX" : VPseudoTernaryWithPolicy<m.wvrclass, GPR, m.vrclass, m,

 multiclass VPseudoVMAC_VV_VX_AAXA<string Constraint = ""> {
   defm "" : VPseudoTernaryV_VV_AAXA<Constraint>,
             Sched<[WriteVIMulAddV, ReadVIMulAddV, ReadVIMulAddV, ReadVIMulAddV, ReadVMask]>;
   defm "" : VPseudoTernaryV_VX_AAXA<Constraint>,
             Sched<[WriteVIMulAddX, ReadVIMulAddV, ReadVIMulAddV, ReadVIMulAddX, ReadVMask]>;
 }
 
 multiclass VPseudoVMAC_VV_VF_AAXA<string Constraint = ""> {
-  defm "" : VPseudoTernaryV_VV_AAXA<Constraint>,
+  defm "" : VPseudoTernaryV_VV_AAXA<Constraint, MxListF>,
             Sched<[WriteVFMulAddV, ReadVFMulAddV, ReadVFMulAddV, ReadVFMulAddV, ReadVMask]>;
   defm "" : VPseudoTernaryV_VF_AAXA<Constraint>,
             Sched<[WriteVFMulAddF, ReadVFMulAddV, ReadVFMulAddV, ReadVFMulAddF, ReadVMask]>;
 }
 
 multiclass VPseudoVSLD_VX_VI<Operand ImmType = simm5, string Constraint = ""> {
   defm "" : VPseudoTernaryV_VX<Constraint>,
             Sched<[WriteVISlideX, ReadVISlideV, ReadVISlideV, ReadVISlideX, ReadVMask]>;

 }
 
 multiclass VPseudoVWMAC_VX {
   defm "" : VPseudoTernaryW_VX,
             Sched<[WriteVIWMulAddX, ReadVIWMulAddV, ReadVIWMulAddV, ReadVIWMulAddX, ReadVMask]>;
 }
 
 multiclass VPseudoVWMAC_VV_VF {
-  defm "" : VPseudoTernaryW_VV,
+  defm "" : VPseudoTernaryW_VV<MxListFW>,
             Sched<[WriteVFWMulAddV, ReadVFWMulAddV, ReadVFWMulAddV, ReadVFWMulAddV, ReadVMask]>;
   defm "" : VPseudoTernaryW_VF,
             Sched<[WriteVFWMulAddF, ReadVFWMulAddV, ReadVFWMulAddV, ReadVFWMulAddF, ReadVMask]>;
 }
 
 multiclass VPseudoVCMPM_VV_VX_VI {
   defm "" : VPseudoBinaryM_VV,
             Sched<[WriteVICmpV, ReadVICmpV, ReadVICmpV, ReadVMask]>;
   defm "" : VPseudoBinaryM_VX,
             Sched<[WriteVICmpX, ReadVICmpV, ReadVICmpX, ReadVMask]>;
   defm "" : VPseudoBinaryM_VI,
             Sched<[WriteVICmpI, ReadVICmpV, ReadVMask]>;
 }
 
 multiclass VPseudoVCMPM_VV_VX {
   defm "" : VPseudoBinaryM_VV,
             Sched<[WriteVICmpV, ReadVICmpV, ReadVICmpV, ReadVMask]>;
   defm "" : VPseudoBinaryM_VX,
             Sched<[WriteVICmpX, ReadVICmpV, ReadVICmpX, ReadVMask]>;
 }
 
 multiclass VPseudoVCMPM_VV_VF {
-  defm "" : VPseudoBinaryM_VV,
+  defm "" : VPseudoBinaryM_VV<MxListF>,
             Sched<[WriteVFCmpV, ReadVFCmpV, ReadVFCmpV, ReadVMask]>;
   defm "" : VPseudoBinaryM_VF,
             Sched<[WriteVFCmpF, ReadVFCmpV, ReadVFCmpF, ReadVMask]>;
 }
 
 multiclass VPseudoVCMPM_VF {
   defm "" : VPseudoBinaryM_VF,
             Sched<[WriteVFCmpF, ReadVFCmpV, ReadVFCmpF, ReadVMask]>;

 multiclass VPseudoVWRED_VS {
   foreach m = MxList in {
     defm _VS : VPseudoTernary<V_M1.vrclass, m.vrclass, V_M1.vrclass, m>,
                Sched<[WriteVIWRedV, ReadVIWRedV, ReadVIWRedV, ReadVIWRedV, ReadVMask]>;
   }
 }
 
 multiclass VPseudoVFRED_VS {
-  foreach m = MxList in {
+  foreach m = MxListF in {
     defm _VS : VPseudoTernary<V_M1.vrclass, m.vrclass, V_M1.vrclass, m>,
                Sched<[WriteVFRedV, ReadVFRedV, ReadVFRedV, ReadVFRedV, ReadVMask]>;
   }
 }
 
 multiclass VPseudoVFREDO_VS {
-  foreach m = MxList in {
+  foreach m = MxListF in {
     defm _VS : VPseudoTernary<V_M1.vrclass, m.vrclass, V_M1.vrclass, m>,
                Sched<[WriteVFRedOV, ReadVFRedOV, ReadVFRedOV, ReadVFRedOV, ReadVMask]>;
   }
 }
 
 multiclass VPseudoVFWRED_VS {
-  foreach m = MxList in {
+  foreach m = MxListF in {
     defm _VS : VPseudoTernary<V_M1.vrclass, m.vrclass, V_M1.vrclass, m>,
                Sched<[WriteVFWRedV, ReadVFWRedV, ReadVFWRedV, ReadVFWRedV, ReadVMask]>;
   }
 }
 
 multiclass VPseudoConversion<VReg RetClass,
                              VReg Op1Class,
                              LMULInfo MInfo,
                              string Constraint = ""> {
   let VLMul = MInfo.value in {
     def "_" # MInfo.MX : VPseudoUnaryNoMask<RetClass, Op1Class, Constraint>;
     def "_" # MInfo.MX # "_MASK" : VPseudoUnaryMaskTA<RetClass, Op1Class,
                                                       Constraint>;
   }
 }
 
 multiclass VPseudoVCVTI_V {
-  foreach m = MxList in
+  foreach m = MxListF in
     defm _V : VPseudoConversion<m.vrclass, m.vrclass, m>,
               Sched<[WriteVFCvtFToIV, ReadVFCvtFToIV, ReadVMask]>;
 }
 
 multiclass VPseudoVCVTF_V {
-  foreach m = MxList in
+  foreach m = MxListF in
     defm _V : VPseudoConversion<m.vrclass, m.vrclass, m>,
               Sched<[WriteVFCvtIToFV, ReadVFCvtIToFV, ReadVMask]>;
 }
 
 multiclass VPseudoConversionW_V {
   defvar constraint = "@earlyclobber $rd";
   foreach m = MxListW in
     defm _V : VPseudoConversion<m.wvrclass, m.vrclass, m, constraint>;
 }
 
 multiclass VPseudoVWCVTI_V {
   defvar constraint = "@earlyclobber $rd";
-  foreach m = MxListW in
+  foreach m = MxListFW in
     defm _V : VPseudoConversion<m.wvrclass, m.vrclass, m, constraint>,
               Sched<[WriteVFWCvtFToIV, ReadVFWCvtFToIV, ReadVMask]>;
 }
 
 multiclass VPseudoVWCVTF_V {
   defvar constraint = "@earlyclobber $rd";
   foreach m = MxListW in
     defm _V : VPseudoConversion<m.wvrclass, m.vrclass, m, constraint>,
               Sched<[WriteVFWCvtIToFV, ReadVFWCvtIToFV, ReadVMask]>;
 }
 
 multiclass VPseudoVWCVTD_V {
   defvar constraint = "@earlyclobber $rd";
-  foreach m = MxListW in
+  foreach m = MxListFW in
     defm _V : VPseudoConversion<m.wvrclass, m.vrclass, m, constraint>,
               Sched<[WriteVFWCvtFToFV, ReadVFWCvtFToFV, ReadVMask]>;
 }
 
 multiclass VPseudoVNCVTI_W {
   defvar constraint = "@earlyclobber $rd";
   foreach m = MxListW in
     defm _W : VPseudoConversion<m.vrclass, m.wvrclass, m, constraint>,
               Sched<[WriteVFNCvtFToIV, ReadVFNCvtFToIV, ReadVMask]>;
 }
 
 multiclass VPseudoVNCVTF_W {
   defvar constraint = "@earlyclobber $rd";
-  foreach m = MxListW in
+  foreach m = MxListFW in
     defm _W : VPseudoConversion<m.vrclass, m.wvrclass, m, constraint>,
               Sched<[WriteVFNCvtIToFV, ReadVFNCvtIToFV, ReadVMask]>;
 }
 
 multiclass VPseudoVNCVTD_W {
   defvar constraint = "@earlyclobber $rd";
-  foreach m = MxListW in
+  foreach m = MxListFW in
     defm _W : VPseudoConversion<m.vrclass, m.wvrclass, m, constraint>,
               Sched<[WriteVFNCvtFToFV, ReadVFNCvtFToFV, ReadVMask]>;
 }
 
 multiclass VPseudoUSSegLoad<bit isFF> {
   foreach eew = EEWList in {
     foreach lmul = MxSet<eew>.m in {
       defvar LInfo = lmul.MX;