Re: How to benchmark a modified ZipGSX?

[Rubywand <rubywand@swbell.net>]

Dan Foster writes ...
> 
> The FAQ mentions an utility called SpeedTest but I get the impression it
> reads current setting off parameter (battery) RAM?
 ....

     It uses a timed machine code instruction loop. So, it's giving you speed
in terms of how quickly the code is executed. Within the round-off, the results
seem to be fairly accurate.

     If you want a closer pre-post changes comparison, you can change the BASIC
to eliminate the "INT" in line 5. Changing the divisor from 10 to 10.5 may
slightly improve accuracy.

     For some changes, the check is unlikely to notice any change. For
instance, a larger cache or more efficient use of cache would probably have no
effect because the code that's running is so small. Timing something like a
search and modify through a large Applesoft document or a complicated fill
under Platinum Paint should give a better indicator of improvement.



Rubywand