OpcodeDispatcher: Add a dirty flag for tracking NZCV status

Cached NZCV reads don't need to be written back at the end of the block.
This will remove one instruction from the end of some blocks.

FEXCore/Source/Interface/Core/OpcodeDispatcher.h

@@ -869,8 +869,9 @@ private:
     }
   }
 
-  OrderedNode* CachedNZCV = {};
-  uint32_t PossiblySetNZCVBits = 0;
+  OrderedNode* CachedNZCV{};
+  bool NZCVDirty{};
+  uint32_t PossiblySetNZCVBits{};
 
   fextl::map<uint64_t, JumpTargetInfo> JumpTargets;
   bool HandledLock{false};
@@ -1115,11 +1116,13 @@ private:
 
   void SetNZCV(OrderedNode *Value) {
     CachedNZCV = Value;
+    NZCVDirty = true;
   }
 
   void ZeroNZCV() {
     CachedNZCV = _Constant(0);
     PossiblySetNZCVBits = 0;
+    NZCVDirty = true;
   }
 
   void ZeroCV() {
@@ -1152,6 +1155,7 @@ private:
     // Mask off just the N bit, which now equals the sign bit
     CachedNZCV = _And(OpSize::i32Bit, Shifted, _Constant(1u << NBit));
     PossiblySetNZCVBits = (1u << NBit);
+    NZCVDirty = true;
   }
 
   void SetNZ_ZeroCV(unsigned SrcSize, OrderedNode *Res) {
@@ -1160,6 +1164,7 @@ private:
     if (SrcSize >= 4) {
       CachedNZCV = _TestNZ(SrcSize, Res);
       PossiblySetNZCVBits = (1u << 31) | (1u << 30);
+      NZCVDirty = true;
     } else {
       // N
       SetN_ZeroZCV(SrcSize, Res);

FEXCore/Source/Interface/Core/OpcodeDispatcher/Flags.cpp

@@ -272,10 +272,11 @@ void OpDispatchBuilder::CalculateAF(OpSize OpSize, OrderedNode *Res, OrderedNode
 void OpDispatchBuilder::CalculateDeferredFlags(uint32_t FlagsToCalculateMask) {
   if (CurrentDeferredFlags.Type == FlagsGenerationType::TYPE_NONE) {
     // Nothing to do
-    if (CachedNZCV)
+    if (NZCVDirty && CachedNZCV)
       _StoreFlag(CachedNZCV, FEXCore::X86State::RFLAG_NZCV_LOC);
 
     CachedNZCV = nullptr;
+    NZCVDirty = false;
     return;
   }
 
@@ -459,10 +460,11 @@ void OpDispatchBuilder::CalculateDeferredFlags(uint32_t FlagsToCalculateMask) {
   // Done calculating
   CurrentDeferredFlags.Type = FlagsGenerationType::TYPE_NONE;
 
-  if (CachedNZCV)
+  if (NZCVDirty && CachedNZCV)
     _StoreFlag(CachedNZCV, FEXCore::X86State::RFLAG_NZCV_LOC);
 
   CachedNZCV = nullptr;
+  NZCVDirty = false;
 }
 
 void OpDispatchBuilder::CalculateFlags_ADC(uint8_t SrcSize, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2, OrderedNode *CF) {