Hi Yasin, As I said, it is about the whole execution plan. To say you can have the different type of joins, more execution steps like filtering, ... I wiped out the excution plan of your two SQL'a and cannot remember fully, but they were different. CPU time is not only and always about buffer gets and sorts. In many cases it is, but for short SQL's buffer gets can be the minority of overall work (you can do around 10,000 simple buffer gets per 100MHz CPU). It may not even be covered with other statistics differences you have got. Show us all major STAT differences, please? Regards, Zoran --- Yasin Baskan <yasbs@xxxxxxxxxxxxxx> wrote: > I have the following stats about sorts. > > NAME > RUN1 RUN2 > DIFF > -------------------------------------------------- > ---------- ---------- > ---------- > STAT...sorts (rows) > 183031 182958 > -73 > > RUN1 is also worse in sorts. I get the query to get > lower number of > blocks but i get an increase in other areas (pysical > reads, latches, > sorts). So the elapsed time increases. > > __________________________________ Do you Yahoo!? All your favorites on one personal page ? Try My Yahoo! http://my.yahoo.com -- //www.freelists.org/webpage/oracle-l