clob datatype is causing performace issue - oracle

Question

UPDATED: The code is working as expected but the performance is very slow. When I do a search without including CLOB data then the query runs very fast but if I include CLOB variable in my search the query is very slow. I am using CLOB to pass large string data('aaaaaaa,bbbb,c,ddddd...') and store those data in global table for better performance, I thought doing such will maximize query performance. How can I improve/utilize my CLOB variable for better perfomance? Please look at the code below for more information. Appreciated for any help. I am still struggling with performance can anyone help/provide any suggestions please.
GLOBAL TT GlobalTemp_EMP( //this already exists
emp_refno (30 byte);
)
Create or replace PROCEDURE Employee(
emp_refno IN CLOB
)
AS
Begin
OPEN p_resultset FOR
with inputs ( str ) as ( //red error line here
select to_clob(emp_refno )
from dual
),
prep ( s, n, token, st_pos, end_pos ) as (
select ',' || str || ',', -1, null, null, 1
from inputs
union all
select s, n+1, substr(s, st_pos, end_pos - st_pos),
end_pos + 1, instr(s, ',', 1, n+3)
from prep
where end_pos != 0
)
INSERT into GlobalTemp_EMP //red error line here
select token from prep;
select e.empname, e.empaddress, f.department
from employee e
join department f on e.emp_id = t.emp_id
and e.emp_refno in (SELECT emp_refno from GlobalTemp_EMP) //using GTT In subquery

Answer 1

put this code between BEGIN and OPEN p_resultset FOR : this might have some performance issue though.
INSERT into GlobalTemp_EMP
with inputs ( str ) as (
select to_clob(emp_refno )
from dual
),
prep ( s, n, token, st_pos, end_pos ) as (
select ',' || str || ',', -1, null, null, 1
from inputs
union all
select s, n+1, substr(s, st_pos, end_pos - st_pos),
end_pos + 1, instr(s, ',', 1, n+3)
from prep
where end_pos != 0
)
select token from prep where token is not NULL;

Answer 2

The below doesn't seem to be valid syntax:
GLOBAL TT GlobalTemp_EMP( //this already exists
emp_refno (30 byte);
)
I don't know the reason for using byte semantics, or whether you defined it as a clob or char or varchar2.
If it is currently a clob, then perhaps you could define the column as emp_refno varchar2(30 char) and add a unique index, changing the Employee procedure to only insert new IDs. An index would help the insertions more than when you read it out.
If you want to insert a huge amount of data into GlobalTemp_EMP faster, I would recommend making it a regular table, pre-processing the data (such as in Perl or other language) to split IDs outside Oracle and then use SQL*Loader. Or perhaps an external table.

Answer 3

I don't think using a global temporary table will improve your performance at all (at least without indexes). Are you sure these are CLOBs? At a glance, these seem to be varchars.
To compare CLOBs, you should be using dbms_lob.compare. I think = will do an implicit conversion to a varchar (and truncate), then do the comparison.