Reduce the sync footprint slightly with less indirection.

apu/bapu/dsp/dsp.cpp

@@ -7,19 +7,6 @@ DSP dsp;
 
 #include "SPC_DSP.cpp"
 
-void DSP::step(unsigned clocks) {
-  clock += clocks;
-}
-
-void DSP::synchronize_smp() {
-  while(clock >= 0) smp.enter();
-}
-
-void DSP::enter() {
-  spc_dsp.run(1);
-  step(24);
-}
-
 uint8 DSP::read(uint8 addr) {
   return spc_dsp.read(addr);
 }

apu/bapu/dsp/dsp.hpp

@@ -3,13 +3,11 @@
 class DSP : public Processor {
 public:
   enum { Threaded = false };
-  alwaysinline void step(unsigned clocks);
   alwaysinline void synchronize_smp();
 
   uint8 read(uint8 addr);
   void write(uint8 addr, uint8 data);
 
-  void enter();
   void power();
   void reset();
 

apu/bapu/smp/core.cpp

@@ -5,11 +5,12 @@ void SMP::tick() {
 
 #ifndef SNES9X
   clock += cycle_step_cpu;
-#else
-  clock++;
-#endif
   dsp.clock -= 24;
   synchronize_dsp();
+#else
+  clock++;
+  dsp.spc_dsp.run(1);
+#endif
 }
 
 void SMP::op_io() {
@@ -77,11 +78,13 @@ void SMP::op_step() {
 
 #ifndef SNES9X
   clock += cycle_table_cpu[opcode];
-#else
-  clock += cycle_count_table[opcode];
-#endif
   dsp.clock -= cycle_table_dsp[opcode];
   synchronize_dsp();
+#else
+  clock += cycle_count_table[opcode];
+  dsp.spc_dsp.run(cycle_count_table[opcode]);
+#endif
+
 
   #endif
 }

apu/bapu/smp/smp.cpp

@@ -34,11 +34,13 @@ void SMP::synchronize_cpu() {
 }
 
 void SMP::synchronize_dsp() {
+#ifndef SNES9X
   if(DSP::Threaded == true) {
   //if(dsp.clock < 0 && scheduler.sync != Scheduler::SynchronizeMode::All) co_switch(dsp.thread);
   } else {
     while(dsp.clock < 0) dsp.enter();
   }
+#endif
 }
 
 void SMP::enter() {