Jit: Skip redundant flushes

This makes codegen faster (by perhaps 10-20% in the case of Jit64,
I didn't measure too closely), which helps speed up NBA Live 2005
a little. But the game still has serious performance issues.

2 files changed, 4 insertions, 4 deletions

diff --git a/Source/Core/Core/PowerPC/Jit64/Jit.cpp b/Source/Core/Core/PowerPC/Jit64/Jit.cpp
index 903bc826fb..f43d25855f 100644
--- a/Source/Core/Core/PowerPC/Jit64/Jit.cpp
+++ b/Source/Core/Core/PowerPC/Jit64/Jit.cpp
@@ -1184,8 +1184,8 @@ bool Jit64::DoJit(u32 em_address, JitBlock* b, u32 nextPC)
         gpr.Discard(op.gprDiscardable);
         fpr.Discard(op.fprDiscardable);
       }
-      gpr.Flush(~op.gprInUse);
-      fpr.Flush(~op.fprInUse);
+      gpr.Flush(~op.gprInUse & (op.regsIn | op.regsOut));
+      fpr.Flush(~op.fprInUse & (op.fregsIn | op.GetFregsOut()));
 
       if (opinfo->flags & FL_LOADSTORE)
         ++js.numLoadStoreInst;
diff --git a/Source/Core/Core/PowerPC/JitArm64/Jit.cpp b/Source/Core/Core/PowerPC/JitArm64/Jit.cpp
index 113c8e0043..3be81d2c71 100644
--- a/Source/Core/Core/PowerPC/JitArm64/Jit.cpp
+++ b/Source/Core/Core/PowerPC/JitArm64/Jit.cpp
@@ -971,8 +971,8 @@ bool JitArm64::DoJit(u32 em_address, JitBlock* b, u32 nextPC)
         gpr.DiscardRegisters(op.gprDiscardable);
         fpr.DiscardRegisters(op.fprDiscardable);
       }
-      gpr.StoreRegisters(~op.gprInUse);
-      fpr.StoreRegisters(~op.fprInUse);
+      gpr.StoreRegisters(~op.gprInUse & (op.regsIn | op.regsOut));
+      fpr.StoreRegisters(~op.fprInUse & (op.fregsIn | op.GetFregsOut()));
 
       if (opinfo->flags & FL_LOADSTORE)
         ++js.numLoadStoreInst;