Fuse icmp/fcmp with select

src/IceTargetLoweringX86Base.h

 //===- subzero/src/IceTargetLoweringX86Base.h - x86 lowering ----*- C++ -*-===//
 //
 //                        The Subzero Code Generator
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 ///
 /// \file
@@ skipping 329 matching lines @@
   }
   void _addps(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Addps::create(Func, Dest, Src0));
   }
   void _addss(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Addss::create(Func, Dest, Src0));
   }
   void _and(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::And::create(Func, Dest, Src0));
   }
+  void _andnps(Variable *Dest, Operand *Src0) {
+    Context.insert(Traits::Insts::Andnps::create(Func, Dest, Src0));
+  }
+  void _andps(Variable *Dest, Operand *Src0) {
+    Context.insert(Traits::Insts::Andps::create(Func, Dest, Src0));
+  }
   void _and_rmw(typename Traits::X86OperandMem *DestSrc0, Operand *Src1) {
     Context.insert(Traits::Insts::AndRMW::create(Func, DestSrc0, Src1));
   }
   void _blendvps(Variable *Dest, Operand *Src0, Operand *Src1) {
     Context.insert(Traits::Insts::Blendvps::create(Func, Dest, Src0, Src1));
   }
   void _br(typename Traits::Cond::BrCond Condition, CfgNode *TargetTrue,
            CfgNode *TargetFalse) {
     Context.insert(Traits::Insts::Br::create(
         Func, TargetTrue, TargetFalse, Condition, Traits::Insts::Br::Far));
@@ skipping 121 matching lines @@
     Inst *NewInst = Traits::Insts::Mov::create(Func, Dest, Src0);
     NewInst->setDestRedefined();
     Context.insert(NewInst);
   }
   void _movd(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Movd::create(Func, Dest, Src0));
   }
   void _movp(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Movp::create(Func, Dest, Src0));
   }
+  void _movp_redefined(Variable *Dest, Operand *Src0) {
+    Inst *NewInst = Traits::Insts::Movp::create(Func, Dest, Src0);
+    NewInst->setDestRedefined();
+    Context.insert(NewInst);
+  }
   void _movq(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Movq::create(Func, Dest, Src0));
   }
   void _movss(Variable *Dest, Variable *Src0) {
     Context.insert(Traits::Insts::MovssRegs::create(Func, Dest, Src0));
   }
   void _movsx(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Movsx::create(Func, Dest, Src0));
   }
   void _movzx(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Movzx::create(Func, Dest, Src0));
   }
+  void _maxss(Variable *Dest, Operand *Src0) {
+    Context.insert(Traits::Insts::Maxss::create(Func, Dest, Src0));
+  }
+  void _minss(Variable *Dest, Operand *Src0) {
+    Context.insert(Traits::Insts::Minss::create(Func, Dest, Src0));
+  }
   void _mul(Variable *Dest, Variable *Src0, Operand *Src1) {
     Context.insert(Traits::Insts::Mul::create(Func, Dest, Src0, Src1));
   }
   void _mulps(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Mulps::create(Func, Dest, Src0));
   }
   void _mulss(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Mulss::create(Func, Dest, Src0));
   }
   void _neg(Variable *SrcDest) {
     Context.insert(Traits::Insts::Neg::create(Func, SrcDest));
   }
   void _nop(SizeT Variant) {
     Context.insert(Traits::Insts::Nop::create(Func, Variant));
   }
   void _or(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Or::create(Func, Dest, Src0));
   }
+  void _orps(Variable *Dest, Operand *Src0) {
+    Context.insert(Traits::Insts::Orps::create(Func, Dest, Src0));
+  }
   void _or_rmw(typename Traits::X86OperandMem *DestSrc0, Operand *Src1) {
     Context.insert(Traits::Insts::OrRMW::create(Func, DestSrc0, Src1));
   }
   void _padd(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Padd::create(Func, Dest, Src0));
   }
   void _pand(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Pand::create(Func, Dest, Src0));
   }
   void _pandn(Variable *Dest, Operand *Src0) {
@@ skipping 125 matching lines @@
     // The xchg modifies Dest and Src -- model that update with a
     // FakeDef/FakeUse.
     Context.insert(
         InstFakeDef::create(Func, Src, llvm::dyn_cast<Variable>(Dest)));
     _set_dest_redefined();
     Context.insert(InstFakeUse::create(Func, Src));
   }
   void _xor(Variable *Dest, Operand *Src0) {
     Context.insert(Traits::Insts::Xor::create(Func, Dest, Src0));
   }
+  void _xorps(Variable *Dest, Operand *Src0) {
+    Context.insert(Traits::Insts::Xorps::create(Func, Dest, Src0));
+  }
   void _xor_rmw(typename Traits::X86OperandMem *DestSrc0, Operand *Src1) {
     Context.insert(Traits::Insts::XorRMW::create(Func, DestSrc0, Src1));
   }
 
   void _iaca_start() {
     if (!BuildDefs::minimal())
       Context.insert(Traits::Insts::IacaStart::create(Func));
   }
   void _iaca_end() {
     if (!BuildDefs::minimal())
@@ skipping 69 matching lines @@
 
   template <typename... Args>
   void dispatchToConcrete(void (Machine::*Method)(Args...), Args &&... args) {
     (static_cast<Machine *>(this)->*Method)(std::forward<Args>(args)...);
   }
 
   void lowerShift64(InstArithmetic::OpKind Op, Operand *Src0Lo, Operand *Src0Hi,
                     Operand *Src1Lo, Variable *DestLo, Variable *DestHi);
 
   /// Emit the code for a combined operation and branch, or set the destination
   /// variable of the operation if Br == nullptr.

[John, 2015/12/07 13:07:44]

The comment needs to be updated.

[sehr, 2015/12/15 20:45:44]

Done.

-  void lowerIcmpAndBr(const InstIcmp *Icmp, const InstBr *Br);
-  void lowerFcmpAndBr(const InstFcmp *Fcmp, const InstBr *Br);
-  void lowerArithAndBr(const InstArithmetic *Arith, const InstBr *Br);
+  void lowerIcmpAndConsumer(const InstIcmp *Icmp, const Inst *Consumer);
+  void lowerFcmpAndConsumer(const InstFcmp *Fcmp, const Inst *Consumer);
+  void lowerArithAndConsumer(const InstArithmetic *Arith, const Inst *Consumer);
 
-  /// Emit a setcc instruction if Br == nullptr; otherwise emit a branch.
-  void setccOrBr(typename Traits::Cond::BrCond Condition, Variable *Dest,
-                 const InstBr *Br);
+  /// Emit a setcc instruction if Consumer == nullptr; otherwise emit a
+  /// specialized version of Consumer.
+  void setccOrConsumer(typename Traits::Cond::BrCond Condition, Variable *Dest,
+                       const Inst *Consumer);
 
-  /// Emit a mov [1|0] instruction if Br == nullptr; otherwise emit a branch.
-  void movOrBr(bool IcmpResult, Variable *Dest, const InstBr *Br);
+  /// Emit a mov [1|0] instruction if Consumer == nullptr; otherwise emit a
+  /// specialized version of Consumer.
+  void movOrConsumer(bool IcmpResult, Variable *Dest, const Inst *Consumer);
+
+  /// Emit the code for instructions with a vector type.
+  void lowerIcmpVector(const InstIcmp *Icmp);
+  void lowerFcmpVector(const InstFcmp *Icmp);
+  void lowerSelectVector(const InstSelect *Inst);
+
+  /// Helpers for select lowering.
+  void lowerSelectMove(Variable *Dest, typename Traits::Cond::BrCond Cond,
+                       Operand *SrcT, Operand *SrcF);
+  void lowerSelectIntMove(Variable *Dest, typename Traits::Cond::BrCond Cond,
+                          Operand *SrcT, Operand *SrcF);
+  /// Generic helper to move an arbitrary type from Src to Dest.
+  void lowerMove(Variable *Dest, Operand *Src, bool IsRedefinition);
+
+  /// Optimizations for idiom recognition.
+  bool lowerOptimizeFcmpSelect(const InstFcmp *Fcmp, const InstSelect *Select);
 
   /// Complains loudly if invoked because the cpu can handle 64-bit types
   /// natively.
   template <typename T = Traits>
   typename std::enable_if<T::Is64Bit, void>::type lowerIcmp64(const InstIcmp *,
-                                                              const InstBr *) {
+                                                              const Inst *) {
     llvm::report_fatal_error(
         "Hey, yo! This is x86-64. Watcha doin'? (lowerIcmp64)");
   }
   /// x86lowerIcmp64 handles 64-bit icmp lowering.
   template <typename T = Traits>
   typename std::enable_if<!T::Is64Bit, void>::type
-  lowerIcmp64(const InstIcmp *Icmp, const InstBr *Br);
+  lowerIcmp64(const InstIcmp *Icmp, const Inst *Consumer);
 
   BoolFolding FoldingInfo;
 };
 } // end of namespace X86Internal
 } // end of namespace Ice
 
 #include "IceTargetLoweringX86BaseImpl.h"
 
 #endif // SUBZERO_SRC_ICETARGETLOWERINGX86BASE_H