Performance fixed. Guessing it's within 30% or so of optimal.

src/opcodes.jl

@@ -47,8 +47,12 @@ export op!, tick!
 #===================================================================================================
     <opcodes>
 ===================================================================================================#
-# returns func, nbytes, ncycles
-const OPCODES = Dict{UInt8,Tuple{Function,Int,Int}}();  sizehint!(OPCODES, N_OPCODES)
+# this is an (efficient) function pointer type for the instructions
+# note this works even though ops take AbstractVector
+const OpFunc = FunctionWrapper{UInt8,Tuple{Chipset,Vector{UInt8}}}
+
+# for the time being opcode 0x00 would throw an error
+const OPCODES = Vector{Tuple{OpFunc,Int,Int}}(N_OPCODES)
 
 # format is
 #    mode, opcode, nbytes, ncycles

test/opcodes.jl

@@ -18,34 +18,39 @@ function makechipset()
 
     cs.ram[0x00] = 0x01
     cs.ram[0x01] = 0x02
-    
+
     cs.ram[0xa000] = 0x05
     cs.ram[0xa001] = 0x06
 
     # @p Sim6502.op0xa5!(cs, [0x00])
     # @p Sim6502.op0xad!(cs, [0x01, 0xa0])
     # @p Sim6502.op0xaa!(cs, UInt8[])
-    @program [0xa5, 0x00, 0xad, 0x01, 0xa0, 0xaa, 0xaa, 0xaa]
+    @program [0xaa, 0xa5, 0x00, 0xad, 0x01, 0xa0, 0xaa, 0xaa, 0xaa]
 
     cs
 end
 
-ref = @benchmarkable begin
-    # cs.cpu.A = 0x00
-    # Sim6502.checkNflag!(cs.cpu, cs.cpu.A)
-    # Sim6502.checkZflag!(cs.cpu, cs.cpu.A)
-    # cs.cpu.A = cs.ram[0xa001]
-    cs.cpu.X = cs.cpu.A
-    Sim6502.checkNflag!(cs.cpu, cs.cpu.X)
-    Sim6502.checkZflag!(cs.cpu, cs.cpu.X)
-end setup=(cs = makechipset())
-
 
-b = @benchmarkable begin
-    # tick!(cs)
-    # tick!(cs)
-    # tick!(cs)
-    tax!(cs.cpu)
-end setup=(cs = makechipset())
+function makebenches()
+    ref = @benchmarkable begin
+        # cs.cpu.A = 0x00
+        # Sim6502.checkNflag!(cs.cpu, cs.cpu.A)
+        # Sim6502.checkZflag!(cs.cpu, cs.cpu.A)
+        # cs.cpu.A = cs.ram[0xa001]
+        cs.cpu.X = cs.cpu.A
+        Sim6502.checkNflag!(cs.cpu, cs.cpu.X)
+        Sim6502.checkZflag!(cs.cpu, cs.cpu.X)
+    end setup=(cs = makechipset())
+
+    b = @benchmarkable begin
+        tick!(cs)
+        tick!(cs)
+        tick!(cs)
+        tick!(cs)
+    end setup=(cs = makechipset())
+
+    ref, b
+end
 
 
+ref, b = makebenches()