measuring performance
__________________________
Previous versions of the simulator did a single step
every idle event (or so)
Here we test the performance change using multiple steps.

c/s : E-cycles per second (as appearing in the About Box)
kc/s : thousands of E-cycles per second
n-step : the effective number of steps for every call to single_step()

*** as a reference, 
	real 68HC11 works at a maximum of 2000 kc/s ( 8 MHz clock)
***

=========== VERSION 1.1.6 =============
Quadra 610 / 8 MB RAM
file: hexmon40.s19
 brief tests on 1-step execution
- with debugger, without cache, 1-step exec.	 330 c/s
- stand-alone, " , "							 640 c/s
- stand-alone with cache, 1-step				1930 c/s

 multi-step tests (on a period of 30 secs)
- 100 steps							 91 kc/s
- 500 steps							107 kc/s
- 1000 steps						159 kc/s
- 2000 steps						167 kc/s
- 10000+ steps 					    178 kc/s

 same as above, but in background (just Think C and a clock)
- 10000 steps					 	116 kc/s
							
============ VERSION 1.2d ===============
Quadra 610
file: hexmon40.s19
 1-step execution
- stand-alone (Think C), 1-step		2670 c/s
- stand-alone (MW CW),   1-step		2420 c/s

=========== VERSION 1.3d ================
Performa 6320 / 16 MB RAM
stand-alone version, compiled with MW CW , profiling, executing with interrupts disabled
file: hexmon40.s19 (on a period of 20 secs)
 1-step execution
- 68k,	1-step		2830 c/s
- PPC,	1-step		2750 c/s

 multi-step, default: 1024 steps
- PPC,	1024 steps		 25 kc/s

same as above, without profiling
 multi-step
- 68k,	1024 steps		106 kc/s
- PPC,	1024 steps		980 kc/s   (almost half the speed of a real 68HC11)