Date   : Wed, 08 Mar 2006 16:16:47 -0000
From   : "David Hunt" <dm.hunt@...>
Subject: Re: Warning: Sad case on list!

> A minor patch.to go a bit quicker:
> 
> 110         A=ZR*ZR:B=ZI*ZI:L=A+B:ZI=2*ZR*ZI+CI:ZR=A-B+CR:I%=I%+1
>                       ^^^^^     ^^^^^    ^^^^
> 120       UNTIL L>4 OR I%>15
>                           ^^^

I tried this change, the replacement of L=SQR(A+B) seems to change the
output considerably. I know if I were to attack this in assembler the SQRT
function would inevitably lead me to a juicy lookup table. I googled and saw
references to "Fast Integer Square Root", Dr. Dobb's Journal, #246, April
1996, or perhaps the iterative Newtonian method. Mind you, it doesn't help
me find a job!

> I don't know how you're doing on your 32016, (the power supply in my home
> cased one had just barfed), so I'm running on the turbo. I think there
> wasn't much in it time wise, especially using BAS32F.

Nope, the 32016 is dawdling along, there was a strange smell in the air
earlier, the 32016 was hot to the touch, so I turned it off (a lucky escape
for me and the NatSemi).

If you remove the outer loop and set E%=1, I get the 'brot up in around 28
mins, otherwise it's 1h10m for all the other iterations too - although it is
nice to see it build, gives you a carrot to watch it, perhaps come back in
10 mins to see how it's doing. Got to try this on the Turbo, bet the Turbo
will be almost on it. Heck, why can't we have a 200MHz 6502 ? WDC do a 20MHz
at 5V - a C-MOS part, it might get upwards of 40MHz at 15V before a bright
flash and puff of smoke signals its final NMI...

> Thanks for the program, I'm going to have to dig out the rest of my notes
> and rebuild the rig!

Good luck with it.

Dave ;)