serial out is dead easy if you think before you act.

I know you do 32*256, but for sample I show 8 high sign, you just make 4 serial threads, one for each 8 rows

the sign dats is stored in memory as one byte per column

the shift registers are wired this way on each module

in > r1 ^->r1 ^-> m  r  ->r1 ^-> out to next module
     r2 ^  r2 ^   o  o    r2 ^  
     r3 ^  r3 ^   r  w    r3 ^  
     r4 ^  r4 ^   e  s    r4 ^  
     r5 ^  r5 ^           r5 ^
     r6 ^  r6 ^           r6 ^
     r7-^  r7-^           r7-^

all you need is

big loop
get char
loop for 8
set bit
clock high
clock low
rotate
if no 8 back to loop for 8
if not all rows
back to big loop

Erik