with t as (select 1000000 q from dual union all
select 200000 from dual union all
select 30000 from dual union all
select 30000 from dual union all -- обратите внимание
select 4000 from dual union all
select 500 from dual union all
select 60 from dual union all
select 7 from dual )
-----------------------
select sq 
from(
  select q, sum(q) over(order by q desc, rownum) sq
       , ceil(.3*count(*) over()) - row_number() over( order by q desc, rownum) f
  from t
) where f = 0

SQ                                      
--------------------------------------- 
1230000                                 

1 rows selected

-- Или как-нибудь так:
with t as (select 1000000 q from dual union all...)
-----------------------
select sq 
from t
model return updated rows
dimension by (ceil(.3*count(*) over()) - row_number() over( order by q desc, rownum) pd)
measures (q sq)
rules(sq[0] = sum(sq)[pd >= 0])

SQ                                      
--------------------------------------- 
1230000                                 

1 rows selected

SQL> with t as (select 1000000 q from dual union all
...
  8  )
  9  select sum(q) from (
 10    select q from t
 11    order by q desc
 12    fetch first 30 percent rows only);

    SUM(Q)
----------
   1230000

SQL>