Using the same method you described (mul/mod) I created the following
qscan:
empty equ dect-10
bomb dat >-86 , <2667
magic equ 5927;3241
magic2 equ 7799;4240
offset equ 2581;7085
off dat offset , offset
dat #26 , #26
dect dat #3 , #20
for 30
dat 0,0
rof
qscan sne comp+2*magic-offset , comp+2*magic2-offset
seq comp+2*magic , comp+2*magic2
jmp dec , }dect
sne comp+5*magic-offset , comp+5*magic2-offset
seq comp+5*magic , comp+5*magic2
jmp dec , <dect
sne comp+8*magic-offset , comp+8*magic2-offset
seq comp+8*magic , comp+8*magic2
jmp dec
sne comp+11*magic-offset , comp+11*magic2-offset
seq comp+11*magic , comp+11*magic2
jmp dec , >dect
sne comp+14*magic-offset , comp+14*magic2-offset
seq comp+14*magic , comp+14*magic2
jmp dec , {dect
sne comp+4*magic-offset , comp+4*magic2-offset
seq comp+4*magic , comp+4*magic2
jmp dec-1 , }dect
sne comp+10*magic-offset , comp+10*magic2-offset
seq comp+10*magic , comp+10*magic2
jmp dec-1 , <dect
sne comp+16*magic-offset , comp+16*magic2-offset
seq comp+16*magic , comp+16*magic2
jmp dec-1
sne comp+22*magic-offset , comp+22*magic2-offset
seq comp+22*magic , comp+22*magic2
jmp dec-1 , >dect
sne comp+28*magic-offset , comp+28*magic2-offset
seq comp+28*magic , comp+28*magic2
jmp dec-1 , {dect
sne comp+24*magic-offset , comp+24*magic2-offset
seq comp+24*magic , comp+24*magic2
jmp }triple , >triple
jmp #0
triple dat #2 , #2
dat 0 , 0
double mul.x -2 , comp
dec mul.x dect , dect
mod.f dect-1 , dect
mul.f dect , comp
comp sne magic , magic2 ;find row
sub.f off , comp
comp2 sne empty , @[EMAIL PROTECTED]
;find a/b-field
mov.x comp , comp
add.b comp , att ;set attack
mov.i bomb , @[EMAIL PROTECTED]
*comp-att
sub #-7 , att
djn.b -3 , #20
jmp #0
end qscan
I think its pretty neat, but the problem is the decoder... just too
large and slow... But the qscan is very easy to extend, instead of
doing the triple-trick I do in the middle, just add more scans jumping
to dec-2, and doing a triple-multiply.
I'm still pondering about a way to make the decoder smaller..
|
