[AArch64][SVE] Implement additional floating-point arithmetic intrinsics

Summary:
Adds intrinsics for the following:
  - ftssel
  - fcadd, fcmla
  - fmla, fmls, fnmla, fnmls
  - fmad, fmsb, fnmad, fnmsb

Reviewers: sdesmalen, huntergr, dancgr, mgudim

Reviewed By: sdesmalen

Subscribers: tschuett, kristof.beyls, hiraditya, rkruppe, psnobl, cameron.mcinally, cfe-commits, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D69707

3 files changed, 104 insertions, 38 deletions

diff --git a/llvm/lib/Target/AArch64/AArch64InstrFormats.td b/llvm/lib/Target/AArch64/AArch64InstrFormats.td
index 17f5277af9e..d6bf9bcd805 100644
--- a/llvm/lib/Target/AArch64/AArch64InstrFormats.td
+++ b/llvm/lib/Target/AArch64/AArch64InstrFormats.td
@@ -758,6 +758,13 @@ def imm0_7 : Operand<i64>, ImmLeaf<i64, [{
   let ParserMatchClass = Imm0_7Operand;
 }
 
+// imm32_0_7 predicate - True if the 32-bit immediate is in the range [0,7]
+def imm32_0_7 : Operand<i32>, ImmLeaf<i32, [{
+  return ((uint32_t)Imm) < 8;
+}]> {
+  let ParserMatchClass = Imm0_7Operand;
+}
+
 // imm32_0_15 predicate - True if the 32-bit immediate is in the range [0,15]
 def imm32_0_15 : Operand<i32>, ImmLeaf<i32, [{
   return ((uint32_t)Imm) < 16;
@@ -10056,15 +10063,20 @@ class ComplexRotationOperand<int Angle, int Remainder, string Type>
   let DiagnosticType = "InvalidComplexRotation" # Type;
   let Name = "ComplexRotation" # Type;
 }
-def complexrotateop : Operand<i32> {
+def complexrotateop : Operand<i32>, ImmLeaf<i32, [{ return Imm >= 0 && Imm <= 270;  }],
+                                                 SDNodeXForm<imm, [{
+  return CurDAG->getTargetConstant((N->getSExtValue() / 90), SDLoc(N), MVT::i32);
+}]>> {
   let ParserMatchClass = ComplexRotationOperand<90, 0, "Even">;
   let PrintMethod = "printComplexRotationOp<90, 0>";
 }
-def complexrotateopodd : Operand<i32> {
+def complexrotateopodd : Operand<i32>, ImmLeaf<i32, [{ return Imm >= 0 && Imm <= 270;  }],
+                                                 SDNodeXForm<imm, [{
+  return CurDAG->getTargetConstant(((N->getSExtValue() - 90) / 180), SDLoc(N), MVT::i32);
+}]>> {
   let ParserMatchClass = ComplexRotationOperand<180, 90, "Odd">;
   let PrintMethod = "printComplexRotationOp<180, 90>";
 }
-
 let mayLoad = 0, mayStore = 0, hasSideEffects = 0 in
 class BaseSIMDThreeSameVectorComplex<bit Q, bit U, bits<2> size, bits<3> opcode,
                                      RegisterOperand regtype, Operand rottype,
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index a732c6e1e31..b2f7cfb9da4 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -145,28 +145,28 @@ let Predicates = [HasSVE] in {
   defm FRECPS_ZZZ  : sve_fp_3op_u_zd<0b110, "frecps",  int_aarch64_sve_frecps_x>;
   defm FRSQRTS_ZZZ : sve_fp_3op_u_zd<0b111, "frsqrts", int_aarch64_sve_frsqrts_x>;
 
-  defm FTSSEL_ZZZ : sve_int_bin_cons_misc_0_b<"ftssel">;
+  defm FTSSEL_ZZZ : sve_int_bin_cons_misc_0_b<"ftssel", int_aarch64_sve_ftssel_x>;
 
-  defm FCADD_ZPmZ : sve_fp_fcadd<"fcadd">;
-  defm FCMLA_ZPmZZ : sve_fp_fcmla<"fcmla">;
+  defm FCADD_ZPmZ : sve_fp_fcadd<"fcadd", int_aarch64_sve_fcadd>;
+  defm FCMLA_ZPmZZ : sve_fp_fcmla<"fcmla", int_aarch64_sve_fcmla>;
 
-  defm FMLA_ZPmZZ  : sve_fp_3op_p_zds_a<0b00, "fmla">;
-  defm FMLS_ZPmZZ  : sve_fp_3op_p_zds_a<0b01, "fmls">;
-  defm FNMLA_ZPmZZ : sve_fp_3op_p_zds_a<0b10, "fnmla">;
-  defm FNMLS_ZPmZZ : sve_fp_3op_p_zds_a<0b11, "fnmls">;
+  defm FMLA_ZPmZZ  : sve_fp_3op_p_zds_a<0b00, "fmla",  int_aarch64_sve_fmla>;
+  defm FMLS_ZPmZZ  : sve_fp_3op_p_zds_a<0b01, "fmls",  int_aarch64_sve_fmls>;
+  defm FNMLA_ZPmZZ : sve_fp_3op_p_zds_a<0b10, "fnmla", int_aarch64_sve_fnmla>;
+  defm FNMLS_ZPmZZ : sve_fp_3op_p_zds_a<0b11, "fnmls", int_aarch64_sve_fnmls>;
 
-  defm FMAD_ZPmZZ  : sve_fp_3op_p_zds_b<0b00, "fmad">;
-  defm FMSB_ZPmZZ  : sve_fp_3op_p_zds_b<0b01, "fmsb">;
-  defm FNMAD_ZPmZZ : sve_fp_3op_p_zds_b<0b10, "fnmad">;
-  defm FNMSB_ZPmZZ : sve_fp_3op_p_zds_b<0b11, "fnmsb">;
+  defm FMAD_ZPmZZ  : sve_fp_3op_p_zds_b<0b00, "fmad",  int_aarch64_sve_fmad>;
+  defm FMSB_ZPmZZ  : sve_fp_3op_p_zds_b<0b01, "fmsb",  int_aarch64_sve_fmsb>;
+  defm FNMAD_ZPmZZ : sve_fp_3op_p_zds_b<0b10, "fnmad", int_aarch64_sve_fnmad>;
+  defm FNMSB_ZPmZZ : sve_fp_3op_p_zds_b<0b11, "fnmsb", int_aarch64_sve_fnmsb>;
 
-  defm FTMAD_ZZI : sve_fp_ftmad<"ftmad">;
+  defm FTMAD_ZZI : sve_fp_ftmad<"ftmad", int_aarch64_sve_ftmad_x>;
 
-  defm FMLA_ZZZI : sve_fp_fma_by_indexed_elem<0b0, "fmla">;
-  defm FMLS_ZZZI : sve_fp_fma_by_indexed_elem<0b1, "fmls">;
+  defm FMLA_ZZZI : sve_fp_fma_by_indexed_elem<0b0, "fmla", int_aarch64_sve_fmla_lane>;
+  defm FMLS_ZZZI : sve_fp_fma_by_indexed_elem<0b1, "fmls", int_aarch64_sve_fmls_lane>;
 
-  defm FCMLA_ZZZI : sve_fp_fcmla_by_indexed_elem<"fcmla">;
-  defm FMUL_ZZZI   : sve_fp_fmul_by_indexed_elem<"fmul">;
+  defm FCMLA_ZZZI : sve_fp_fcmla_by_indexed_elem<"fcmla", int_aarch64_sve_fcmla_lane>;
+  defm FMUL_ZZZI   : sve_fp_fmul_by_indexed_elem<"fmul", int_aarch64_sve_fmul_lane>;
 
   // SVE floating point reductions.
   defm FADDA_VPZ   : sve_fp_2op_p_vd<0b000, "fadda">;
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 467b573627f..05c396f78bf 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -299,7 +299,8 @@ class SVE_3_Op_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
       (inst $Op1, $Op2, $Op3)>;
 
 class SVE_4_Op_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
-                   ValueType vt2, ValueType vt3, ValueType vt4, Instruction inst>
+                   ValueType vt2, ValueType vt3, ValueType vt4,
+                   Instruction inst>
 : Pat<(vtd (op vt1:$Op1, vt2:$Op2, vt3:$Op3, vt4:$Op4)),
       (inst $Op1, $Op2, $Op3, $Op4)>;
 
@@ -1225,7 +1226,7 @@ multiclass sve_fp_2op_p_zds_fscale<bits<4> opc, string asm,
 }
 
 class sve_fp_ftmad<bits<2> sz, string asm, ZPRRegOp zprty>
-: I<(outs zprty:$Zdn), (ins zprty:$_Zdn, zprty:$Zm, imm0_7:$imm3),
+: I<(outs zprty:$Zdn), (ins zprty:$_Zdn, zprty:$Zm, imm32_0_7:$imm3),
   asm, "\t$Zdn, $_Zdn, $Zm, $imm3",
   "",
   []>, Sched<[]> {
@@ -1245,10 +1246,17 @@ class sve_fp_ftmad<bits<2> sz, string asm, ZPRRegOp zprty>
   let ElementSize = ElementSizeNone;
 }
 
-multiclass sve_fp_ftmad<string asm> {
+multiclass sve_fp_ftmad<string asm, SDPatternOperator op> {
   def _H : sve_fp_ftmad<0b01, asm, ZPR16>;
   def _S : sve_fp_ftmad<0b10, asm, ZPR32>;
   def _D : sve_fp_ftmad<0b11, asm, ZPR64>;
+
+  def : Pat<(nxv8f16 (op (nxv8f16 ZPR16:$Zn), (nxv8f16 ZPR16:$Zm), (i32 imm32_0_7:$imm))),
+            (!cast<Instruction>(NAME # _H) ZPR16:$Zn, ZPR16:$Zm, imm32_0_7:$imm)>;
+  def : Pat<(nxv4f32 (op (nxv4f32 ZPR32:$Zn), (nxv4f32 ZPR32:$Zm), (i32 imm32_0_7:$imm))),
+            (!cast<Instruction>(NAME # _S) ZPR32:$Zn, ZPR32:$Zm, imm32_0_7:$imm)>;
+  def : Pat<(nxv2f64 (op (nxv2f64 ZPR64:$Zn), (nxv2f64 ZPR64:$Zm), (i32 imm32_0_7:$imm))),
+            (!cast<Instruction>(NAME # _D) ZPR64:$Zn, ZPR64:$Zm, imm32_0_7:$imm)>;
 }
 
 
@@ -1323,10 +1331,14 @@ class sve_fp_3op_p_zds_a<bits<2> sz, bits<2> opc, string asm, ZPRRegOp zprty>
   let ElementSize = zprty.ElementSize;
 }
 
-multiclass sve_fp_3op_p_zds_a<bits<2> opc, string asm> {
+multiclass sve_fp_3op_p_zds_a<bits<2> opc, string asm, SDPatternOperator op> {
   def _H : sve_fp_3op_p_zds_a<0b01, opc, asm, ZPR16>;
   def _S : sve_fp_3op_p_zds_a<0b10, opc, asm, ZPR32>;
   def _D : sve_fp_3op_p_zds_a<0b11, opc, asm, ZPR64>;
+
+  def : SVE_4_Op_Pat<nxv8f16, op, nxv8i1, nxv8f16, nxv8f16, nxv8f16, !cast<Instruction>(NAME # _H)>;
+  def : SVE_4_Op_Pat<nxv4f32, op, nxv4i1, nxv4f32, nxv4f32, nxv4f32, !cast<Instruction>(NAME # _S)>;
+  def : SVE_4_Op_Pat<nxv2f64, op, nxv2i1, nxv2f64, nxv2f64, nxv2f64, !cast<Instruction>(NAME # _D)>;
 }
 
 class sve_fp_3op_p_zds_b<bits<2> sz, bits<2> opc, string asm,
@@ -1354,10 +1366,14 @@ class sve_fp_3op_p_zds_b<bits<2> sz, bits<2> opc, string asm,
   let ElementSize = zprty.ElementSize;
 }
 
-multiclass sve_fp_3op_p_zds_b<bits<2> opc, string asm> {
+multiclass sve_fp_3op_p_zds_b<bits<2> opc, string asm, SDPatternOperator op> {
   def _H : sve_fp_3op_p_zds_b<0b01, opc, asm, ZPR16>;
   def _S : sve_fp_3op_p_zds_b<0b10, opc, asm, ZPR32>;
   def _D : sve_fp_3op_p_zds_b<0b11, opc, asm, ZPR64>;
+
+  def : SVE_4_Op_Pat<nxv8f16, op, nxv8i1, nxv8f16, nxv8f16, nxv8f16, !cast<Instruction>(NAME # _H)>;
+  def : SVE_4_Op_Pat<nxv4f32, op, nxv4i1, nxv4f32, nxv4f32, nxv4f32, !cast<Instruction>(NAME # _S)>;
+  def : SVE_4_Op_Pat<nxv2f64, op, nxv2i1, nxv2f64, nxv2f64, nxv2f64, !cast<Instruction>(NAME # _D)>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -1384,26 +1400,34 @@ class sve_fp_fma_by_indexed_elem<bits<2> sz, bit opc, string asm,
   let ElementSize = ElementSizeNone;
 }
 
-multiclass sve_fp_fma_by_indexed_elem<bit opc, string asm> {
-  def _H : sve_fp_fma_by_indexed_elem<{0, ?}, opc, asm, ZPR16, ZPR3b16, VectorIndexH> {
+multiclass sve_fp_fma_by_indexed_elem<bit opc, string asm,
+                                      SDPatternOperator op> {
+  def _H : sve_fp_fma_by_indexed_elem<{0, ?}, opc, asm, ZPR16, ZPR3b16, VectorIndexH32b> {
     bits<3> Zm;
     bits<3> iop;
     let Inst{22} = iop{2};
     let Inst{20-19} = iop{1-0};
     let Inst{18-16} = Zm;
   }
-  def _S : sve_fp_fma_by_indexed_elem<0b10, opc, asm, ZPR32, ZPR3b32, VectorIndexS> {
+  def _S : sve_fp_fma_by_indexed_elem<0b10, opc, asm, ZPR32, ZPR3b32, VectorIndexS32b> {
     bits<3> Zm;
     bits<2> iop;
     let Inst{20-19} = iop;
     let Inst{18-16} = Zm;
   }
-  def _D : sve_fp_fma_by_indexed_elem<0b11, opc, asm, ZPR64, ZPR4b64, VectorIndexD> {
+  def _D : sve_fp_fma_by_indexed_elem<0b11, opc, asm, ZPR64, ZPR4b64, VectorIndexD32b> {
     bits<4> Zm;
     bit iop;
     let Inst{20} = iop;
     let Inst{19-16} = Zm;
   }
+
+  def : Pat<(nxv8f16 (op nxv8f16:$Op1, nxv8f16:$Op2, nxv8f16:$Op3, (i32 VectorIndexH32b:$idx))),
+            (!cast<Instruction>(NAME # _H) $Op1, $Op2, $Op3, VectorIndexH32b:$idx)>;
+  def : Pat<(nxv4f32 (op nxv4f32:$Op1, nxv4f32:$Op2, nxv4f32:$Op3, (i32 VectorIndexS32b:$idx))),
+            (!cast<Instruction>(NAME # _S) $Op1, $Op2, $Op3, VectorIndexS32b:$idx)>;
+  def : Pat<(nxv2f64 (op nxv2f64:$Op1, nxv2f64:$Op2, nxv2f64:$Op3, (i32 VectorIndexD32b:$idx))),
+            (!cast<Instruction>(NAME # _D) $Op1, $Op2, $Op3, VectorIndexD32b:$idx)>;
 }
 
 
@@ -1425,26 +1449,33 @@ class sve_fp_fmul_by_indexed_elem<bits<2> sz, string asm, ZPRRegOp zprty,
   let Inst{4-0}   = Zd;
 }
 
-multiclass sve_fp_fmul_by_indexed_elem<string asm> {
-  def _H : sve_fp_fmul_by_indexed_elem<{0, ?}, asm, ZPR16, ZPR3b16, VectorIndexH> {
+multiclass sve_fp_fmul_by_indexed_elem<string asm, SDPatternOperator op> {
+  def _H : sve_fp_fmul_by_indexed_elem<{0, ?}, asm, ZPR16, ZPR3b16, VectorIndexH32b> {
     bits<3> Zm;
     bits<3> iop;
     let Inst{22} = iop{2};
     let Inst{20-19} = iop{1-0};
     let Inst{18-16} = Zm;
   }
-  def _S : sve_fp_fmul_by_indexed_elem<0b10, asm, ZPR32, ZPR3b32, VectorIndexS> {
+  def _S : sve_fp_fmul_by_indexed_elem<0b10, asm, ZPR32, ZPR3b32, VectorIndexS32b> {
     bits<3> Zm;
     bits<2> iop;
     let Inst{20-19} = iop;
     let Inst{18-16} = Zm;
   }
-  def _D : sve_fp_fmul_by_indexed_elem<0b11, asm, ZPR64, ZPR4b64, VectorIndexD> {
+  def _D : sve_fp_fmul_by_indexed_elem<0b11, asm, ZPR64, ZPR4b64, VectorIndexD32b> {
     bits<4> Zm;
     bit iop;
     let Inst{20} = iop;
     let Inst{19-16} = Zm;
   }
+
+  def : Pat<(nxv8f16 (op nxv8f16:$Op1, nxv8f16:$Op2, (i32 VectorIndexH32b:$idx))),
+            (!cast<Instruction>(NAME # _H) $Op1, $Op2, VectorIndexH32b:$idx)>;
+  def : Pat<(nxv4f32 (op nxv4f32:$Op1, nxv4f32:$Op2, (i32 VectorIndexS32b:$idx))),
+            (!cast<Instruction>(NAME # _S) $Op1, $Op2, VectorIndexS32b:$idx)>;
+  def : Pat<(nxv2f64 (op nxv2f64:$Op1, nxv2f64:$Op2, (i32 VectorIndexD32b:$idx))),
+            (!cast<Instruction>(NAME # _D) $Op1, $Op2, VectorIndexD32b:$idx)>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -1476,10 +1507,17 @@ class sve_fp_fcmla<bits<2> sz, string asm, ZPRRegOp zprty>
   let ElementSize = zprty.ElementSize;
 }
 
-multiclass sve_fp_fcmla<string asm> {
+multiclass sve_fp_fcmla<string asm, SDPatternOperator op> {
   def _H : sve_fp_fcmla<0b01, asm, ZPR16>;
   def _S : sve_fp_fcmla<0b10, asm, ZPR32>;
   def _D : sve_fp_fcmla<0b11, asm, ZPR64>;
+
+  def : Pat<(nxv8f16 (op nxv8i1:$Op1, nxv8f16:$Op2, nxv8f16:$Op3, nxv8f16:$Op4, (i32 complexrotateop:$imm))),
+            (!cast<Instruction>(NAME # _H) $Op1, $Op2, $Op3, $Op4, complexrotateop:$imm)>;
+  def : Pat<(nxv4f32 (op nxv4i1:$Op1, nxv4f32:$Op2, nxv4f32:$Op3, nxv4f32:$Op4, (i32 complexrotateop:$imm))),
+            (!cast<Instruction>(NAME # _S) $Op1, $Op2, $Op3, $Op4, complexrotateop:$imm)>;
+  def : Pat<(nxv2f64 (op nxv2i1:$Op1, nxv2f64:$Op2, nxv2f64:$Op3, nxv2f64:$Op4, (i32 complexrotateop:$imm))),
+            (!cast<Instruction>(NAME # _D) $Op1, $Op2, $Op3, $Op4, complexrotateop:$imm)>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -1509,19 +1547,24 @@ class sve_fp_fcmla_by_indexed_elem<bits<2> sz, string asm,
   let ElementSize = ElementSizeNone;
 }
 
-multiclass sve_fp_fcmla_by_indexed_elem<string asm> {
-  def _H : sve_fp_fcmla_by_indexed_elem<0b10, asm, ZPR16, ZPR3b16, VectorIndexS> {
+multiclass sve_fp_fcmla_by_indexed_elem<string asm, SDPatternOperator op> {
+  def _H : sve_fp_fcmla_by_indexed_elem<0b10, asm, ZPR16, ZPR3b16, VectorIndexS32b> {
     bits<3> Zm;
     bits<2> iop;
     let Inst{20-19} = iop;
     let Inst{18-16} = Zm;
   }
-  def _S : sve_fp_fcmla_by_indexed_elem<0b11, asm, ZPR32, ZPR4b32, VectorIndexD> {
+  def _S : sve_fp_fcmla_by_indexed_elem<0b11, asm, ZPR32, ZPR4b32, VectorIndexD32b> {
     bits<4> Zm;
     bits<1> iop;
     let Inst{20} = iop;
     let Inst{19-16} = Zm;
   }
+
+  def : Pat<(nxv8f16 (op nxv8f16:$Op1, nxv8f16:$Op2, nxv8f16:$Op3, (i32 VectorIndexS32b:$idx), (i32 complexrotateop:$imm))),
+            (!cast<Instruction>(NAME # _H) $Op1, $Op2, $Op3, VectorIndexS32b:$idx, complexrotateop:$imm)>;
+  def : Pat<(nxv4f32 (op nxv4f32:$Op1, nxv4f32:$Op2, nxv4f32:$Op3, (i32 VectorIndexD32b:$idx), (i32 complexrotateop:$imm))),
+            (!cast<Instruction>(NAME # _S) $Op1, $Op2, $Op3, VectorIndexD32b:$idx, complexrotateop:$imm)>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -1552,10 +1595,17 @@ class sve_fp_fcadd<bits<2> sz, string asm, ZPRRegOp zprty>
   let ElementSize = zprty.ElementSize;
 }
 
-multiclass sve_fp_fcadd<string asm> {
+multiclass sve_fp_fcadd<string asm, SDPatternOperator op> {
   def _H : sve_fp_fcadd<0b01, asm, ZPR16>;
   def _S : sve_fp_fcadd<0b10, asm, ZPR32>;
   def _D : sve_fp_fcadd<0b11, asm, ZPR64>;
+
+  def : Pat<(nxv8f16 (op nxv8i1:$Op1, nxv8f16:$Op2, nxv8f16:$Op3, (i32 complexrotateopodd:$imm))),
+            (!cast<Instruction>(NAME # _H) $Op1, $Op2, $Op3, complexrotateopodd:$imm)>;
+  def : Pat<(nxv4f32 (op nxv4i1:$Op1, nxv4f32:$Op2, nxv4f32:$Op3, (i32 complexrotateopodd:$imm))),
+            (!cast<Instruction>(NAME # _S) $Op1, $Op2, $Op3, complexrotateopodd:$imm)>;
+  def : Pat<(nxv2f64 (op nxv2i1:$Op1, nxv2f64:$Op2, nxv2f64:$Op3, (i32 complexrotateopodd:$imm))),
+            (!cast<Instruction>(NAME # _D) $Op1, $Op2, $Op3, complexrotateopodd:$imm)>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -5646,10 +5696,14 @@ class sve_int_bin_cons_misc_0_b<bits<2> sz, string asm, ZPRRegOp zprty>
   let Inst{4-0}   = Zd;
 }
 
-multiclass sve_int_bin_cons_misc_0_b<string asm> {
+multiclass sve_int_bin_cons_misc_0_b<string asm, SDPatternOperator op> {
   def _H : sve_int_bin_cons_misc_0_b<0b01, asm, ZPR16>;
   def _S : sve_int_bin_cons_misc_0_b<0b10, asm, ZPR32>;
   def _D : sve_int_bin_cons_misc_0_b<0b11, asm, ZPR64>;
+
+  def : SVE_2_Op_Pat<nxv8f16, op, nxv8f16, nxv8i16, !cast<Instruction>(NAME # _H)>;
+  def : SVE_2_Op_Pat<nxv4f32, op, nxv4f32, nxv4i32, !cast<Instruction>(NAME # _S)>;
+  def : SVE_2_Op_Pat<nxv2f64, op, nxv2f64, nxv2i64, !cast<Instruction>(NAME # _D)>;
 }
 
 class sve_int_bin_cons_misc_0_c<bits<8> opc, string asm, ZPRRegOp zprty>