PEAK optimisation. 420mips. it's not getting much faster tbh.

dsa/emulator/src/lib.rs

@@ -1,4 +1,5 @@
 #![feature(likely_unlikely)]
+#![feature(inherent_associated_types)]
 
 pub mod args;
 pub mod config;

dsa/emulator/src/processor/processor.rs

@@ -158,7 +158,7 @@ impl<Mem: RandomAccessMemory> Emulator<Mem> {
     #[cold]
     fn boot(&mut self) -> Result<(), String> {
         if !(self.__io_configured && self.__mmap_configured) {
-            return Err("Processor not configured".to_string());
+            return Err("Emulator<Mem> not configured".to_string());
         }
 
         if let Some(map) = &self.memory_map {
@@ -207,6 +207,10 @@ impl<Mem: RandomAccessMemory> Emulator<Mem> {
         let mut time = Instant::now();
 
         'emu: loop {
+            // IMPORTANT
+            // do not change anything about this loop. it's fully optimised afaik.
+            // Check for commands or hardware Interrupts every 32k cycles
+            if unlikely(self.internal_state.clock & 0x7FFF == 0) {
                 // Update UI thread (roughly every 512k cycles targeting 60 UPS)
                 if unlikely(self.internal_state.clock & 0x7FFFF == 0) {
                     if self.shared_state.update_req.load(Ordering::Relaxed) {
@@ -214,9 +218,7 @@ impl<Mem: RandomAccessMemory> Emulator<Mem> {
                     }
                 }
 
-            // Check for commands or hardware Interrupts every 32k cycles
-            if self.internal_state.clock % 0x7FFF == 0 {
-                if self.shared_state.running.load(Ordering::Relaxed) == false {
+                if unlikely(!self.shared_state.running.load(Ordering::Relaxed)) {
                     self.idle_wait();
                 }
 
@@ -260,6 +262,7 @@ impl<Mem: RandomAccessMemory> Emulator<Mem> {
             }
 
             self.mut_reg(Register::Pcx).add_assign(4);
+
             self.execute(Instruction(instruction));
 
             // Check if executing the interrupt caused a fault.