arrays - how to optimise intersect on heroku -

i have postgresql query, check length of intersection of 2 arrays. simplified version of query like:

select array_length(users.array & array[1,2,3,4]) users; 

it part of larger query, not important here. works fine on local database, on heroku, intarray extension not whitelisted.

i found simple function intersect arrays, compared & operator rather slow.

create or replace function array_intersect(anyarray, anyarray) returns anyarray $$   select array( select * unnest( $1 ) unnest = any( $2 ) ); $$ language sql; 

on 2000 records of table using & operator takes 50ms , using function above takes 150ms. compare many records can , function doesn't scale '&' operator.

is there way faster or add intarray heroku?

guess depend on rest of query, , how optimized need be.

my first attempt create cte of insecting lengths using unnest , intersect... fetch values cte within main expression needed... along lines of:

with merged (     select * unnest(array[1,2,3])      intersect      select * unnest(array[1]) )  select count(*) length merged; 

if inappropriate have tried function (as did), since interested in length have made function return (to avoid needless conversion array , back):

create or replace function array_count_intersects(anyarray, anyarray) returns bigint $$     select count(*) (         select * unnest($1) unnest = any($2)     ) merg; $$ language sql;  select array_count_intersects(array[1,2,3], array[1]); 

if still not enough, @ creating intermediate table store/cache lengths of intersecting arrays updated via triggers when data changes, use intermediate table lengths need in main query without need perform unions @ all.