DT> There are times when an RTE200 patched program may work at one CPU
 DT> speed level, but not at higher ones.  I'm not sure why that is, but
 DT> I think I've seen it happen a few times.

i've touched on this numerous times in the past in various echos, as well...

the main reason is that one quick patch only increased the size of the
counter but that was not enough when things got really fast... some
increased the size of the counter and didn't take into account that the
calculation routine also needed to have vars resized so what happened is
that the bottom half of the resized was passed but could still let a 0
value in at times... the real fixes generally replace the existing method
of averaging how long it takes the NOP function to execute... the thing to
watch for in these timing loop situations is that an instruction may
execute in 4 clock ticks on this processor but may take longer on a newer
one... that would happen due to the manufacturer optimizing things here and
there... did you know that there are x86 processors that don't ROTR (ROTate
Right) or ROTL (ROTate Left) any more and it has to be emulated which costs
many clock ticks?? this happened why? because they determined that those
instructions were being used all that much...

)\/(ark