It depends on the code you write and optimization levels. Here's some examples to get you started.

1) simple one row select

select col
into x
from MY_TABLE
where ...

Obviously arraysize of no real consequence here, we are getting 1 row

2) simple cursor

for i in ( select * from MY_TABLE ) loop
...
end loop;

We'll do an array fetch size of 100 for you, even though the code doesnt seem to suggest it

3) DML

insert into MY_TABLE select * from OTHER_TABLE

PL/SQL has no 'participation' in this command, so it will run like a normal SQL and arraysize is not of relevance

4)

declare
cursor C is ...
begin
open C;
loop
fetch C into ...
exit when c%notfound;
end loop;

Will do arrayfetch 1 row at a time... probably a bad idea. Better to recast as

for i in C loop

5) explcit

declare
cursor C is ...
begin
open C;
loop
fetch C bulk collect into xxx limit nnn;
exit when c%notfound;
end loop;

arraysize is 'nnn';

6) simple multi row select

select col
bulk collect into my_array
from MY_TABLE
where ...

Arraysize as big as the rows picked up.

Hope this helps.